Publications per Research Domain

Intelligent Systems and Analytics, Tasks and Data Management in Pervasive & Edge Computing and IoT

Koukosias, A., Anagnostopoulos, C., Kolomvatsos, K., ‘Task-Aware Data Selectivity in Pervasive Edge Computing Environments’, IEEE Transactions on Knowledge and Data Engineering (TKDE), 2024.
Abstract

×
Context-aware data selectivity in Edge Computing (EC) requires nodes to efficiently manage the data collected from Internet of Things (IoT) devices, e.g., sensors, for supporting real-time and data-driven pervasive analytics. Data selectivity at the network edge copes with the challenge of deciding which data should be kept at the edge for future analytics tasks under limited computational and storage resources. Our challenge is to efficiently learn the access patterns of data-driven tasks (analytics) and predict which data are relevant, thus, being stored in nodes' local datasets. Task patterns directly indicate which data need to be accessed and processed to support end-users' applications. We introduce a task workload-aware mechanism which adopts one-class classification to learn and predict the relevant data requested by past tasks. The inherent uncertainty in learning task patterns, identifying inliers and eliminating outliers is handled by introducing a lightweight fuzzy inference estimator that dynamically adapts nodes' local data filters ensuring accurate data relevance prediction. We analytically describe our mechanism and comprehensively evaluate and compare against baselines and approaches found in the literature showcasing its applicability in pervasive EC.
Albanis, G., Zioulis, N., Kolomvatsos, K., ‘BudleMoCap++: Efficient, robust and smooth motion capture from sparse Multiview videos’, Computer Vision and Image Understanding, Elsevier, 2024.
Abstract

×
Producing smooth and accurate motions from sparse videos without requiring specialized equipment and markers is a long-standing problem in the research community. Most approaches typically involve complex processes such as temporal constraints, multiple stages combining data-driven regression and optimization techniques, and bundle solving over temporal windows. These increase the computational burden and introduce the challenge of hyperparameter tuning for the different objective terms. In contrast, BundleMoCap++ offers a simple yet effective approach to this problem. It solves the motion in a single stage, eliminating the need for temporal smoothness objectives while still delivering smooth motions without compromising accuracy. BundleMoCap++ outperforms the state-of-the-art without increasing complexity. Our approach is based on manifold interpolation between latent keyframes. By relying on a local manifold smoothness assumption and appropriate interpolation schemes, we efficiently solve a bundle of frames using two or more latent codes. Additionally, the method is implemented as a sliding window optimization and requires only the first frame to be properly initialized, reducing the overall computational burden. BundleMoCap++’s strength lies in achieving high-quality motion capture results with fewer computational resources. To do this efficiently, we propose a novel human pose prior that focuses on the geometric aspect of the latent space, modeling it as a hypersphere, allowing for the introduction of sophisticated interpolation techniques. We also propose an algorithm for optimizing the latent variables directly on the learned manifold, improving convergence and performance. Finally, we introduce high-order interpolation techniques adapted for the hypersphere, allowing us to increase the solving temporal window, enhancing performance and efficiency.
Aladwani, T., Anagnostopoulos, C., Kolomvatsos, K., 'Node and Relevant Data Selection in Distributed Predictive Analytics: A Query-centric Approach', Journal of Network and Computer Applications, Elsevier, 2024.
Abstract

×
Distributed Predictive Analytics (DPA) refers to constructing predictive models based on data distributed across nodes. DPA reduces the need for data centralization, thus, alleviating concerns about data privacy, decreasing the load on central servers, and minimizing communication overhead. However, data collected by nodes are inherently different; each node can have different distributions, volumes, access patterns, and features space. This heterogeneity hinders the development of accurate models in a distributed fashion. Many state-of-the-art methods adopt random node selection as a straightforward approach. Such method is particularly ineffective when dealing with data and access pattern heterogeneity, as it increases the likelihood of selecting nodes with low-quality or irrelevant data for DPA. Consequently, it is only after training models over randomly selected nodes that the most suitable ones can be identified based on the predictive performance. This results in more time and resource consumption, and increased network load. In this work, holistic knowledge of nodes’ data characteristics and access patterns is crucial. Such knowledge enables the successful selection of a subset of suitable nodes for each DPA task (query) before model training. Our method engages the most suitable nodes by predicting their relevant distributed data and learning predictive models per query. We introduce a novel DPA query-centric mechanism for node and relevant data selection. We contribute with (i) predictive selection mechanisms based on the availability and relevance of data per DPA query and (ii) various distributed machine learning mechanisms that engage the most suitable nodes for model training. We evaluate the efficiency of our mechanism and provide a comparative assessment with other methods found in the literature. Our experiments showcase that our mechanism significantly outperforms other approaches being applicable in DPA.
Moustakas, A., Tziouvaras, A., Kolomvatsos, K., 'Data and Resource Aware Incremental ML Training in Support of Pervasive Applications', Computing, Springer, 2024.
Abstract

×
Nowadays, the use of Artificial Intelligence (AI) and Machine Learning (ML) algorithms is increasingly affecting the performance of innovative systems. At the same time, the advent of the Internet of Things (IoT) and the Edge Computing (EC) as means to place computational resources close to users create the need for new models in the training process of ML schemes due to the limited computational capabilities of the devices/nodes placed there. In any case, we should not forget that IoT devices or EC nodes exhibit less capabilities than the Cloud back end that could be adopted for a more complex training upon vast volumes of data. The ideal case is to have, at least, basic training capabilities at the IoT-EC ecosystem in order to reduce the latency and face the needs of near real time applications. In this paper, we are motivated by this need and propose a model that tries to save time in the training process by focusing on the training dataset and its statistical description. We do not dive into the architecture of any ML model as we target to provide a more generic scheme that can be applied upon any ML module. We monitor the statistics of the training dataset and the loss during the process and identify if there is a potential to stop it when not significant contribution is foreseen for the data not yet adopted in the model. We argue that our approach can be applied only when a negligibly decreased accuracy is acceptable by the application gaining time and resources from the training process. We provide two algorithms for applying this approach and an extensive experimental evaluation upon multiple supervised ML models to reveal the benefits of the proposed scheme and its constraints.
Moustakas, A., Kolomvatsos, K., 'Drift-Based Task Management in Support of Pervasive Edge Applications', Internet of Things, Elsevier, 2024.
Abstract

×
The Internet of Things (IoT) creates a sprawling network where multiple devices can interact, enabling a variety of devices to communicate, collect information, and carry out tasks in support of services that end users may enjoy. Edge Computing, known for its lower latency compared to the Cloud, has sparked interest in managing the execution of tasks within a huge ecosystem that acts as a cover upon the IoT infrastructure. The primary challenge lies in maximizing the use of limited edge resources while minimizing response time, prompting numerous research endeavors. However, existing efforts often overlook shifts in the collected data that may affect the execution of tasks and the production of knowledge. This paper focuses on developing a mechanism that considers data and concept drifts to optimize the management of tasks. The ultimate goal is to maximize the accuracy levels while optimizing resource utilization, through a tasks’ offloading scheme that accounts for distribution-based similarity, offering substantial benefits in managing these constrained resources efficiently. The shifts in data can help determine if the execution of a task is efficient in a specific node and become part of a reasoning model that guides the offloading decisions. Finally, we evaluate our model using a large set of experimental scenarios.
Papathanasiou, D., Kolomvatsos, K., 'Data Management & Selectivity in Collaborative Pervasive Edge Computing', Computing, Springer, 2024.
Abstract

×
Context-aware data management becomes the focus of several research efforts, which can be placed at the intersection between the Internet of Things (IoT) and Edge Computing (EC). Huge volumes of data captured by IoT devices are processed in EC environments. Even if edge nodes undertake the responsibility of data management tasks, they are characterized by limited storage and computational resources compared to Cloud. Apparently, this mobilises the introduction of intelligent data selection methods capable of deciding which of the collected data should be kept locally based on end users/applications requests. In this paper, we devise a mechanism where edge nodes learn their own data selection filters and decide the distributed allocation of newly collected data to their peers and/or Cloud once these data are not conformed with the local data filters. Our mechanism intents to postpone final decisions on data transfer to Cloud (e.g., data centers) to pervasively keep relevant data as close and as long to end users/applications as possible. The proposed mechanism derives a data-selection map across edge nodes by learning specific data sub-spaces, which facilitate the placement of processing tasks (e.g., analytics queries). This is very critical when we target to support near real time decision making and would like to minimize all parts of the tasks allocation procedure. We evaluate and compare our approach against baselines and schemes found in the literature showcasing its applicability in pervasive edge computing environments.
Long, Q., Anagnostopoulos, C., Kolomvatsos, K., 'Enhancing Knowledge Reusability: A Distributed Multitask Machine Learning Approach', IEEE Transactions on Emerging Topics in Computing, 2024.
Abstract

×
In the era of the Internet of Things, the unprecedented growth of data surpasses current predictive analytics and processing capabilities. Due to the potential redundancy of similar data and analytics tasks, it is imperative to extract patterns from distributed data and predictive models so that existing schemes can be efficiently reused in distributed computing environments. This is expected to avoid building and maintaining reduplicative predictive models. The fundamental challenge, however, is the detection of reusable tasks and tuning models in order to improve predictive capacity while being reused. We introduce a two-phase Distributed Multi-task Machine Learning (DMtL) framework coping with this challenge. In the first phase, similar tasks are identified and efficiently grouped together according to locally trained models' performance meta-features, using Partial Learning Curves (PLC). In the subsequent phase, we leverage the PLC-driven DMtL paradigm to boost the performance of candidate reusable models per group of tasks in distributed computing environments. We provide a thorough analysis of our framework along with a comparative assessment against relevant approaches and prior work found in the respective literature. Our experimental results showcase the feasibility of the PLC-driven DMtL method in terms of adaptability and reusability of existing knowledge in distributed computing systems.
Lioliopoulos, P., Oikonomou, P., Boulougaris, G., Kolomvatsos, K., 'Integrated Portable and Stationary Health Impact-Monitoring System for Firefighters', Sensors, MDPI, 2024.
Abstract

×
The multi-layered negative effects caused by pollutants released into the atmosphere as a result of fires served as the stimulus for the development of a system that protects the health of firefighters operating in the affected area. A collaborative network comprising mobile and stationary Internet of Things (IoT) devices that are furnished with gas sensors, along with a remote server, constructs a resilient framework that monitors the concentrations of harmful emissions, characterizes the ambient air quality of the vicinity where the fire transpires, adopting European Air Quality levels, and communicates the outcomes via suitable applications (RESTful APIs and visualizations) to the stakeholders responsible for fire management decision making. Different experimental evaluations adopting separate contexts illustrate the operation of the infrastructure.
Kolomvatsos, K., Anagnostopoulos, C., 'Autonomous Proactive Data Management in Support of Pervasive Edge Applications', Future Generation Computer Systems (FGCS), Elsevier, 2024.
Abstract

×
Recently, context-aware data management becomes the focus of many research efforts placed at the intersection between the Internet of Things (IoT) and Edge Computing (EC). Huge volumes of data can be collected by IoT devices being `connected' with EC environments transferring data towards the Cloud. EC nodes undertake the responsibility of managing the collected data, however, they are characterized by limited storage and computational resources compared to Cloud. Evidently, this makes imperative the introduction of data selectivity methods to keep locally only the data requested by end users or applications for current and future analytics services. In this paper, we study an EC environment where nodes rely on data selectivity and decide the allocation of newly received data to peers, or Cloud when these data are not conformed with local data filters. Data filters are the means for determining local data selectivity by keeping only data that statistically match the needs of nodes (e.g., match the already present data or requests for processing defined by incoming tasks). We contribute with data selectivity and filtering models that support intelligent decisions on when and where incoming data should be allocated. We intent to `postpone' the transfer of data to the Cloud by keeping them close to end users. Our approach concludes a data map of an EC environment nominating every node as the owner of specific data (sub)spaces facilitating the placement of future processing tasks. We evaluate and compare our models and algorithms against schemes found in the literature showcasing their applicability and efficiency in pervasive edge computing environments.
Lioliopoulos, P., Oikonomou, P., Boulougaris, G., Kolomvatsos, K., 'Real-Time Monitoring of Wildfire Pollutants for Health Impact Assessment', in IEEE International Geoscience and Remote Sensing Symposium (IGARSS),, July 7-12, Athens, Greece, 2024.
Abstract

×
Recognizing the severe health implications of dangerous pollutants emitted during a wildfire incident, we introduce a robust monitoring framework based on the Internet of Things (IoT) paradigm designed for real-time remote sensing of wildfire pollutants. The system is specifically adapted for measuring the wildfire health impact on firefighters and nearby residents. Its architecture comprises portable and stationary solutions and a Web application, enabling easy access to emission data and Air Quality Index (AQI) for interested parties and command \& control centers. The portable solution benefits firefighters by providing real-time air quality information, aiding in decision-making for safety and efficient firefighting, while the stationary one contributes to mitigating the risk of potential evacuation in a region near the fire incident. Interested parties e.g., local authorities, command centers and environmental monitoring agencies, can easily and seamlessly integrate our system into their operations by initiating simple requests in our REST API.
Kolomvatsos, K., Anagnostopoulos, C., 'Autonomous Proactive Data Management in Support of Pervasive Edge Applications', Future Generation Computer Systems (FGCS), Elsevier, 2024.
Abstract

×
Recently, context-aware data management becomes the focus of many research efforts placed at the intersection between the Internet of Things (IoT) and Edge Computing (EC). Huge volumes of data can be collected by IoT devices being `connected' with EC environments transferring data towards the Cloud. EC nodes undertake the responsibility of managing the collected data, however, they are characterized by limited storage and computational resources compared to Cloud. Evidently, this makes imperative the introduction of data selectivity methods to keep locally only the data requested by end users or applications for current and future analytics services. In this paper, we study an EC environment where nodes rely on data selectivity and decide the allocation of newly received data to peers, or Cloud when these data are not conformed with local data filters. Data filters are the means for determining local data selectivity by keeping only data that statistically match the needs of nodes (e.g., match the already present data or requests for processing defined by incoming tasks). We contribute with data selectivity and filtering models that support intelligent decisions on when and where incoming data should be allocated. We intent to `postpone' the transfer of data to the Cloud by keeping them close to end users. Our approach concludes a data map of an EC environment nominating every node as the owner of specific data (sub)spaces facilitating the placement of future processing tasks. We evaluate and compare our models and algorithms against schemes found in the literature showcasing their applicability and efficiency in pervasive edge computing environments.
ALFahad, S., Wang, Q., Anagnostopoulos, C., Kolomvatsos, K., 'Task Offloading in Mobile Edge Computing using Cost-based Discounted Optimal Stopping', Open Computer Science, 2024.
Abstract

×
Mobile Edge Computing (MEC) paradigm has emerged to improve the Quality of Service & Experience of applications deployed in close proximity to end-users. Due to their restricted computational and communication resources, MEC nodes can provide access to a portion of the entire set of services and data gathered. Therefore, there are several obstacles to their management. By keeping track of all the services offered by the MEC nodes is challenging, particularly if their demand rates change over time. Received tasks (like, analytics queries, classification tasks, and model learning) require services to be invoked in real MEC use-case scenarios, e.g., smart cities. It is not unusual for a node to lack the necessary services or part of them. Undeniably, not all the requested services may be locally available, thus, MEC nodes must deal with the timely and appropriate choice of whether to carry out a service replication (pull-action) or tasks offloading (push-action) to peer nodes in a MEC environment. In this paper, we contribute with a novel time-optimized mechanism based on the Optimal Stopping Theory, which is built on the cost-based decreasing service demand rates evidenced in various service management situations. Our mechanism tries to optimally solve the decision-making dilemma between pull- and push-action. The experimental findings of our mechanism and its comparative assessment with other methods found in the literature showcase the achieved optimal decisions with respect to certain cost-based objective functions over dynamic service demand rates.
Wang, Q., ALFahad, S., Mateo Fornes, J., Anagnostopoulos, C., Kolomvatsos, K., 'Resilient Edge Predictive Analytics by Enhancing Local Models', Open Computer Science, 2024.
Abstract

×
In distributed computing environments, the collaboration of nodes for predictive analytics at the network edge plays a crucial role in supporting real-time services. When a node’s service turns unavailable for various reasons (e.g., service updates, node maintenance, or even node failure), the rest available nodes could not efficiently replace its service due to different data and predictive models (e.g., Machine Learning (ML) models). To address this, we propose decision-making strategies rooted in statistical signatures of nodes’ data. Specifically, these signatures referto the unique patterns and behaviors within each node’s data that can be leveraged to predict the suitability of potential surrogate nodes. Recognizing and acting on these statistical nuances ensures a more targeted and efficient response to node failures. Such strategies aim to identify surrogate nodes capable of substituting for failing nodes’ services by building enhanced predictive models. Our resilient framework helps to guide the task requests from failing nodes to the most appropriate surrogate nodes. In this case, the surrogate nodes can use their enhanced models, which can produce equivalent and satisfactory results for the requested tasks. We provide experimental evaluations and comparative assessments with baseline approaches over real datasets. Our results showcase the capability of our framework to maintain the overall performance of predictive analytics under nodes’ failures in edge computing environments.
Moustakas, T., Kolomvatsos, K., 'Homogeneous Transfer Learning for Supporting Pervasive Edge Applications', Journal of Evolving Systems, Springer, 2023.
Abstract

×
Edge computing is a paradigm which refers to the use of a range of devices and computational capabilities close to end users. The main idea is to process data closer to the location they are being generated, enabling processing with low latency leading to a model that is appropriate to support real time inference. Transfer learning and its usage in edge computing have recently received significant attention by the research community. Many research efforts focus on the improvement of transfer learning and/or exploit it in the edge setting. A major limitation of these efforts is the failure to consider the diverse statistical distributions representing each dimension of a dataset. In this paper, we provide a novel mechanism for identifying multidimensional distribution-based similarity on the available datasets and performing transfer learning according to this similarity. Additionally, our proposed model, Distribution Based Transfer Learning (DBTL), effectively addresses the uncertainty inherent in these setups, which influences the decision-making process for determining the optimal transfer learning and training procedures. We elaborate on the use of fuzzy logic to deliver an uncertainty-driven decision making model and present the outcomes of our experimental evaluation. The ultimate goal is to reveal the advantages of the proposed solution and provide its applicability in real setups.
Moustakas, T., Kolomvatsos, K., 'Cluster based Similarity Extraction upon Distributed Datasets', Cluster Computing Journal, Springer, 2023.
Abstract

×
The Internet of Things (IoT) offers a vast infrastructure where numerous devices interact to collect data or perform simple processing activities. These devices are, usually, equipped with sensors, software and storage capabilities being able to process the collected data and exchange knowledge with their peers. Data and knowledge may be transferred, in an upwards mode, to the edge infrastructure and Cloud to be the subject of more complex processing. Advanced processing may be also realized upon clusters of IoT devices or edge nodes, i.e., a set of interconnected devices/nodes. Such group of nodes may work to achieve a common goal, thus, we may desire to rely on their collected data to support activities like federated learning. A training process, when necessary, may be delivered across the cluster of nodes to conclude the view of the group. In this paper, we focus on the development of a monitoring process upon the collected data and a grouping method upon multiple datasets formulated by IoT devices (hosted by edge nodes) in order to deliver the most similar ones. Our vision is to continuously identify similar datasets to support knowledge extraction through the adoption of techniques like the aforementioned federated learning. We utilize a correlation detection method combined with a probabilistic model to conclude the most similar datasets.
Vrachimis, A., Gkegka, S., Kolomvatsos, K., 'Resilient Edge Machine Learning in Smart City Environments', Journal of Smart Cities and Society, 2023.
Abstract

×
Distributed Machine Learning (DML) has emerged as a disruptive technology that enables the execution of Machine Learning (ML) and Deep Learning (DL) algorithms in proximity to data generation, facilitating predictive analytics services in Smart City environments. However, the real-time analysis of data generated by Smart City Edge Devices (EDs) poses significant challenges. Concept drift, where the statistical properties of data streams change over time, leads to degraded prediction performance. Moreover, the reliability of each computing node directly impacts the availability of DML systems, making them vulnerable to node failures. To address these challenges, we propose a resilience framework comprising computationally lightweight maintenance strategies that ensure continuous quality of service and availability in DML applications. We conducted a comprehensive experimental evaluation using real datasets, assessing the effectiveness and efficiency of our resilience maintenance strategies across three different scenarios. Our findings demonstrate the significance and practicality of our framework in sustaining predictive performance in smart city edge learning environments. Specifically, our enhanced model exhibited increased generalizability when confronted with concept drift. Furthermore, we achieved a substantial reduction in the amount of data transmitted over the network during the maintenance of the enhanced models, while balancing the trade-off between the quality of analytics and inter-node data communication cost.
Moustakas, A., Kolomvatsos, K., 'Correlation Adaptive Task Scheduling', Springer Computing Journal, 2023.
Abstract

×
The Internet of Things (IoT) offers a vast infrastructure where numerous devices interact to collect data or perform processing activities (tasks). These devices are, usually, equipped with sensors, software and storage capabilities being able to process the collected data. Task scheduling in edge computing environments has gained considerable attention lately due to the fact that the Edge Computing provides lower latency compared to the Cloud. The main challenge is to find a way to maximize the utilization of limited resources available in the edge compared to the Cloud and minimize response time. Many research efforts have been published in order to overcome this challenge. The main limitation of these efforts is the fact that they do not account for task requirements or task correlation that originates from these requirements. In this paper, we focus on the development of a mechanism that utilizes correlation between tasks and takes task requirements into consideration in order to provide efficient task scheduling. Our vision is to minimize task failures and maximize resource utilization with great benefits for the efficient management of the limited resources.
Boulougaris, G., Kolomvatsos, K., 'An Inference Mechanism for Proactive Service Migration at the Edge', IEEE Transactions on Network and Service Management, 2023.
Abstract

×
The coexistence of the Internet of Things and Edge Computing aims to offer a processing infrastructure close to end users that will improve the performance of applications and limit the latency in the provision of services. Services are adopted to assist in the execution of tasks imposed by the requests of end users/applications. The implementation of an effective framework for services management in the distributed edge nodes is necessary to achieve the aforementioned goals. The discussed framework ought to address the trade-off between overheads related to services migration/replication and data transmission. In this paper, we propose a proactive statistical model for allocating the available services upon the observed demand and supporting edge nodes to decide when and where it is necessary to migrate/replicate them. Our aim is to place services at locations where an increased demand is observed, however, under the uncertainty about the future evolution of the incoming requests. We elaborate on the evaluation of the proposed model and provide a comparative assessment with relevant schemes adopting real datasets. Our experimental validation demonstrates that our approach reinforces the heterogeneous engaged edge nodes to correctly infer the time instance and the location when/where services should be migrated/replicated to meet the dynamics of their demand. The interesting is that the proposed model achieves encouraging outcomes when it is adopted to cope with the mobility of users.
Wang, Q., Anagnostopoulos, C., Fornes, J., Kolomvatsos, K., Vrachimis, A., 'Maintenance of Model Resilience in Distributed Edge Learning Environments', in 19th International Conference on Intelligent Environments, Island of Mauritius, 27-30 June 2023.
Abstract

×
Distributed Machine Learning (DML) at the edge of the network involves model learning and inference across networking nodes over distributed data. One type of model learning could be the delivery of predictive analytics services to formulate intelligent environments, however, those environments heavily rely on real-time inference and are significantly influenced by changes in the underlying data (concept drifts). Moreover, the quality of service and availability in DML environments are directly tied to each node’s reliability since such environments are highly susceptible to the impact of node failures. Even if such challenges can be tackled with distributed resilience mechanisms, their effectiveness and efficiency, due to concept drifts, should be maintained to ensure continuous and sustained quality of service. DML systems operate in dynamic environments, thus, they require their models to be updated according to the novel trends embedded in the new data they encounter. We, therefore, introduce several model maintenance mechanisms to ensure resilient DML systems in the long term when concept drifts emerge. We provide a comprehensive experimental evaluation of our resilience maintenance mechanisms over synthetic and real data showcasing their importance and applicability in edge learning environments.
Aladwani, T., Anagnostopoulos, C., Kolomvatsos, K., Alghamdi, I., Deligianni, F., 'Query-driven Edge Node Selection in Distributed Learning Environments', in Data-driven Smart Cities (DASC) in 39th IEEE International Conference on Data Engineering (ICDE 2023), Anaheim, California, USA, April 3-7 2023.
Abstract

×
Computing nodes in Edge Computing environments (e.g., vehicles, smartphones, and drones) collect and share unlimited data. Such data are exploited to locally build Machine Learning (ML) models for applications such as predictive analytics, exploratory analysis, and smart applications. This edge node-centric local learning reduces the need for data transfer and centralization, which is affected by different factors such as data privacy, data size, communication overhead, and computing resource limitations. Therefore, a collaborative learning fashion at the network edge has appeared as a promising paradigm that enables multiple distributed (edge) nodes to train and deploy ML models cooperatively without infringement of data privacy. Nevertheless, the variety, distribution and quality of data vary between edge nodes. Hence, evidently, selecting unsuitable edge nodes can have a negative impact on the ML model performances. We have devised (i) an intelligent edge node selection mechanism per analytics query based on the range of the availability of the required training data at the edge and (ii) variants of collaborative learning processes engaging the most suitable (selected) nodes for models training and inference. We evaluate the efficiency of our node selection mechanism and collaborative learning processes and provide a comparative assessment with other methods found in the literature using real data. Our experimental results showcase that the proposed mechanism along with a variety of learning paradigms significantly outperforms baseline approaches and existing node selection mechanisms in distributed computing environments.
Papathanasaki, M., Fountas, P., Kolomvatsos, K., 'An Uncertainty-Driven Proactive Self-Healing Model for Pervasive Applications', Network, MDPI, 2022.
Abstract

×
The ever-increasing demand for services of end-users in the Internet of Things (IoT) often causes great congestion in the nodes dedicated to serving their requests. Such nodes are usually placed at the edge of the network, becoming the intermediates between the IoT infrastructure and Cloud. Edge nodes offer many advantages when adopted to perform processing activities that are realized close to end-users, limiting the latency in the provision of responses. In this article, we attempt to solve the problem of the potential overloading of edge nodes by proposing a mechanism that always keeps free space in their queue to host high-priority processing tasks. We introduce a proactive, self-healing mechanism that utilizes the principles of Fuzzy Logic, in combination with a non-parametric statistical method that reveals the trend of nodes’ loads as depicted by the incoming tasks and their capability to serve them in the minimum possible time. Through our approach, we manage to ensure the uninterrupted service of high-priority tasks, taking into consideration the demand for tasks as well. Based on this approach, we ensure the fastest possible delivery of results to the requestors while keeping the latency for serving high-priority tasks at the lowest possible levels. A set of experimental scenarios is adopted to evaluate the performance of the suggested model by presenting the corresponding numerical results.
Wang, Q., Mateo Fornes, J., Anagnostopoulos, C., Kolomvatsos, K., 'Predictive Model Resilience in Edge Computing', in 8th IEEE World Forum on Internet of Things (IEEE WFIoT2022), Oct. 26 – Nov. 11, Yokohama, Japan, 2022.
Abstract

×
Node failure is a commonly seen threat in distributed Machine Learning systems. It is hard to predict having a huge negative impact on system availability to provide e.g., predictive analytics. Considering the benefits obtained from reduced latency and bandwidth overhead in Edge Computing (EC), invocation of the Cloud should be avoided. Hence, finding the best substitute nodes at the network edge to be invoked %serve invocations of instead of failing nodes, evidently, builds the system's resilience upon node failures. To achieve this goal, we contribute with a resilience mechanism that relies on several data-mixing strategies that build enhanced models in each node. Such models have satisfactory prediction capabilities to handle failing nodes' predictive tasks, thus, ensuring resilience in predictive services. Furthermore, we propose a graph-driven approach to guide node invocation minimising the performance loss upon node failures. Our performance evaluation and comparative assessment showcase the applicability of our model resilience approach in intelligent EC.
Long, Q., Kolomvatsos, K., Anagnostopoulos, C., 'Knowledge Reuse in Edge Computing Environments', Journal of Network and Computer Applications (JNCA), Elsevier, 2022.
Abstract

×
To cope with the challenge of managing numerous computing devices, humongous data volumes and models in Internet-of-Things environments, Edge Computing (EC) has emerged to serve latency-sensitive and compute-intensive applications. Although EC paradigm significantly eliminates latency for predictive analytics tasks by deploying computation on edge nodes' vicinity, the large scale of EC infrastructure still has huge inescapable burdens on the required resources. This paper introduces a novel paradigm where edge nodes effectively reuse local completed computations (e.g., trained models) at the network edge, coined as knowledge reuse. Such paradigm releases the burden from individual nodes, where they can save resources by relying on reusing models for various predictive analytics tasks (e.g., regression and classification). We study the feasibility of our paradigm by involving pair-wise (dis)similarity metrics among datasets over nodes based on statistical learning techniques (kernel-based Maximum Mean Discrepancy and eigenspace Cosine Dissimilarity). Our paradigm is enhanced with computationally lightweight monitoring mechanisms, which rely on Holt-Winters to forecast future violations and updates of the reused models. Such mechanisms predict when 'borrowed' models are insufficient for being reused, triggering a new process of finding more appropriate models to be reused at the network edge. We provide comprehensive performance evaluation and comparative assessment of our algorithms over different experimental scenarios using real and synthetic datasets. Our findings showcase the ability and robustness of our paradigm to maintain up-to-date reused models at the edge trading off quality of analytics and resource utilization.
Kolomvatsos, K., 'A Proactive Inference Scheme for Data-Aware Decision Making in Support of Pervasive Applications', Future Generation Computer Systems (FGCS), Elsevier, 2022.
Abstract

×
The advent of the Internet of Things (IoT) offers a huge infrastructure where numerous devices can collect and process data retrieved by their environment. Due to the limited computational capabilities of IoT devices, the adoption of the Edge Computing (EC) ecosystem can provide an additional layer of processing to offer more computational resources compared to the IoT. In EC, one can find an increased number of nodes that can collaborate each other and, collectively, support advanced processing activities very close to end users enhancing the pervasiveness of services/applications. Usual collaborative activities can be met around the exchange of data or services (e.g., data/services migration) or offloading actions for tasks demanding a specific processing workflow upon the collected data. The collective intelligence of the EC ecosystem should rely on a map of the available nodes and their resources/capabilities in order to support efficient decision making for the aforementioned activities. In this paper, we propose a model that creates this map and proactively infers the matching between EC nodes based on their data. Our inference is based on the temporal probabilistic management of data synopses exchanged between peers in the EC ecosystem while exposing the historical correlation of the individual/distributed datasets. The adoption of a decision making scheme upon synopses can limit the circulation of data in the network and increase the speed of processing. We elaborate on an aggregation scheme applied on the outcomes of a probabilistic model and a correlation analysis scheme presenting and elaborating on the theoretical background of the proposed solution. We experiment upon real datasets and a number of evaluation scenarios to reveal the performance of the approach while placing it in the respective literature through a comparative assessment.
Boulougaris, G., Kolomvatsos, K., 'A QoS-aware, Proactive Tasks Offloading Model for Pervasive Applications', in 9th International Conference on Future Internet of Things and Cloud (FiCloud), 22-24 Aug, Rome, Italy, 2022.
Abstract

×
Edge Computing (EC) is a promising paradigm that provides multiple computation and analytics capabilities close to data sources while alleviating the drawbacks of centralized systems. Nonetheless, due to the limited computational resources of EC nodes and the expectation of ensuring high levels of QoS during tasks execution, innovative task management approaches are required. In this paper, we propose, built on the autonomous character of EC nodes, a distributed and intelligent decision-making scheme for tasks scheduling, considering multiple criteria/parameters. Our aim is to enhance the behavior of EC nodes making them capable of securing high QoS levels during their functioning. Every EC node systematically evaluates the probability of violating the desired QoS levels and proactively decides some tasks to be offloaded to peer nodes or Cloud when the aforementioned condition stands true. We present, describe and evaluate the proposed scheme through multiple experimental scenarios revealing its performance and the benefits of the envisioned monitoring mechanism when serving processing requests in very dynamic environments like the EC.
Aladwani, T., Anagnostopoulos, C., Alghamdi, I., Kolomvatsos, K., 'Data-driven Analytics Task Management at the Edge: A Fuzzy Reasoning Approach', in 9th International Conference on Future Internet of Things and Cloud (FiCloud), 22-24 Aug, Rome, Italy, 2022.
Abstract

×
Dynamic data-driven applications such as tracking and surveillance have emerged in Internet of Things (IoT) environments. Such applications rely heavily on data generated by connected devices (e.g., sensors). Consequently, leveraging these data in building data-driven predictive analytics tasks improves the Quality of Service (QoS) and, as a result, Quality of Experience (QoE). Such data support various data-driven tasks such as regression and classification. Analytics tasks require data and resources to be executed at the edge since transferring them to the cloud negatively affects response times and QoS. However, the network edge is characterized by limited resources compared to the cloud, being the subject of constraints that are violated upon offloading data-driven tasks to improper edge nodes. We contribute with an analytics task management mechanism based on the context of the requested data, the task delay sensitivity and the VM utilization. We introduce a novel Fuzzy inference mechanism for determining whether data-driven tasks should be executed locally, offloaded to peer edge servers, or sent to cloud. We showcase how our fuzzy reasoning mechanism efficiently derives such decisions by calculating the offloading probability per task. The derived optimal actions are compared against benchmark models in Edge Computing (EC) environments.
Scotti, X., Anagnostopoulos, C., Kolomvatsos, K., 'On the Reusability of Machine Learning Models in Edge Computing: A Statistical Learning Approach', in Future Technologies Conference (FTC), 20-21 October, Vancouver, Canada, 2022.
Abstract

×
The adoption of Edge Computing continues to grow with edge nodes recording increasingly more data, which inevitably requires that they should be processed through Machine Learning (ML) models to speed up the production of knowledge. However, training these models requires an increased amount of resources, which are limited, thus, the reuse of models becomes of paramount importance. Given that we do not have a pool of models to choose from, is it possible to determine which nodes in the network require distinct models and which of them could be reused? In this paper, we propose a solution to this question, an online model reuse framework which is evaluated for its precision and speedup. The framework considers all possible combinations of pairs in the network to determine adopting statistical learning methods. Then for each pair, the node model is chosen that has the highest inlier data space overlap. We further propose a decision making algorithm to unify information from the pair level to the network level. Our extensive experimental analysis in the context of both regression and classification shows the feasibility our solution in model reusability in Edge Computing environments.
Anagnostopoulos, C., Aladwani, T., Alghamdi, I., Kolomvatsos, K., 'Data-driven Analytics Task Management Reasoning Mechanism in Edge Computing', Smart Cities, MDPI, 5, 562–582, 2022.
Abstract

×
Internet of Things (IoT) applications have led to exploding contextual data for predictive analytics and exploration tasks. Consequently, computationally data-driven tasks at the network edge, such as machine learning models' training and inference, have become more prevalent. Such tasks require data and resources to be executed at the network edge, while transferring data to Cloud servers negatively affects expected response times and quality of service (QoS). In this paper, we study certain computational offloading techniques in autonomous computing nodes (ANs) at the edge. ANs are distinguished by limited resources that are subject to a variety of constraints that can be violated when executing analytical tasks. In this context, we contribute a task-management mechanism based on approximate fuzzy inference over the popularity of tasks and the percentage of overlapping between the data required by a data-driven task and data available at each AN. Data-driven tasks' popularity and data availability are fed into a novel two-stages Fuzzy Logic (FL) inference system that determines the probability of either executing tasks locally, offloading them to peer ANs or offloading to Cloud. We showcase that our mechanism efficiently derives such probability per each task, which consequently leads to efficient uncertainty management and optimal actions compared to benchmark models.
Fountas, P., Papathanasaki, M., Kolomvatsos, K., Anagnostopoulos, C., 'Query Driven Data Subspace Mapping', in IFIP 18th International Conference on Artificial Intelligence Applications and Innovations (AIAI), Crete, Greece, Hybrid, 17 20 June, 2022.
Abstract

×
The increased use of multiple types of computer systems and smart devices in several areas, has created massive amounts of data. Concurrently, the need for a subset of these data by numerous applications and users for task execution and knowledge extraction, has resulted in the injection of a massive number of queries per second into distributed database servers. As a result of this phenomena, a major process is the efficient response of these queries both in terms of time and the detection of acceptable data, while rejecting undesired data points. In this paper, we present a hierarchical query-driven clustering approach, for performing efficient data mapping in remote databases for future incoming queries. We distinguish ourselves from current methods, by combining the technique of Query-Based Learning (QBL) with a hierarchical clustering of multiple forms of clustering in the same model. The suggested model's performance is assessed, using a number of experimental scenarios as well as numerical data.
Harth, N., Anagnostopoulos, C., Voegel, H.-J., Kolomvatsos, K., 'Local & Federated Learning at the Network Edge for Efficient Predictive Analytics', Future Generation Computer Systems, Elsevier, 2022.
Abstract

×
The ability to perform computation on devices present in the Internet of Things (IoT) and Edge Computing (EC) environments leads to bandwidth, storage, and energy constraints, as most of these devices are limited with resources. Using such device computational capacity, coined as Edge Devices (EDs), in performing locally Machine Learning (ML) and analytics tasks enables accurate and real-time predictions at the network edge. The locally generated data in EDs is contextual and, for resource efficiency reasons, should not be distributed over the network. In such context, the local trained models need to adapt to occurring concept drifts and potential data distribution changes to guarantee a high prediction accuracy. We address the importance of personalization and generalization in EDs to adapt to data distribution over evolving environments. In the following work, we propose a methodology that relies on Federated Learning (FL) principles to ensure the generalization capability of the locally trained ML models. Moreover, we extend FL with Optimal Stopping Theory (OST) and adaptive weighting over personalized and generalized models to incorporate optimal model selection decision making. We contribute with a personalized, efficient learning methodology in EC environments that can swiftly select and switch models inside the EDs to provide accurate predictions towards changing environments. Theoretical analysis of the optimality and uniqueness of the proposed solution is provided. Additionally, comprehensive comparative and performance evaluation over real contextual data streams testing our methodology against current approaches in the literature for FL and centralized learning are provided concerning information loss and prediction accuracy metrics. We showcase improvement of the prediction quality towards FL-based approaches by at least 50% using our methodology.
Kolomvatsos, K., Anagnostopoulos, C., 'A Proactive Statistical Model Supporting Services and Tasks Management in Pervasive Applications', IEEE Transactions on Network and Service Management, 2022.
Abstract

×
The combination of the Internet of Things (IoT) and Edge Computing (EC) can support intelligent pervasive applications that meet the needs of end users. A challenge is to provide efficient inference models for supporting collaborative activities. EC nodes can interact with IoT devices and each other to conclude those activities producing knowledge. In this paper, we propose a proactive scheme to decide upon the efficient management of services and tasks present/reported at EC nodes. Services can be processing modules applied upon local data. We monitor the demand for the available services and reason upon their management, i.e., for their local presence/invocation as the demand is updated by the requested processing activities. For each incoming task, an inference process is fired to proactively meet the strategic targets of the envisioned model. We propose a statistical inference process upon the demand for services and the contextual performance data of nodes combining it with a utility aware decision making model. Instead of exclusively focusing on services migration or tasks offloading as other relevant efforts do, we elaborate on the decision making for the selection of one of the aforementioned activities (the most appropriate at a specific time instance). We present our model and evaluate it through a high number of simulations to expose its pros and cons placing it in the respective literature as one of the first attempts to proactively decide the presence of services to an ecosystem of processing nodes.
Fountas, P., Kolomvatsos, K., 'An Ensemble Model for Data Imputation to Support Pervasive Applications', International Journal on Artificial Intelligence Tools, 2022.
Abstract

×
Pervasive Computing (PC) opens up the room for the adoption of devices very close to end users that gives the opportunity to interact with them and execute various applications to facilitate their every day activities. Pervasive applications are supported by the evolution of the Internet of Things (IoT) as well as the Edge Computing (EC) that offer vast infrastructures where data can be collected and processed. EC acts as the mediator between the IoT and the Cloud becoming the middle point where data are transferred before they become the subject of processing by Cloud services. IoT devices can assist in the collection of data and EC nodes could play the role of intermediate processing points executing the desired tasks requested by applications. Any processing can be affected by the presence of missing values that may jeopardize the quality of the outcomes. In this paper, we propose a setting where EC nodes play the aforementioned processing role for data reported by IoT devices and adopt an ensemble scheme for data imputation in the case where missing values are present. Our model relies on the local view of the IoT devices reporting a data vector with a missing value, the view of the group that consists of the IoT nodes with a high similarity in the reported data and a probabilistic approach that reveals the statistics of data as realized in the group of similar reports. The proposed scheme continuously detects the correlation between the incoming data streams and efficiently combines the available data vectors before it is in a position to suggest replacements for missing values. The envisioned aggregation mechanism is capable of resulting the appropriate replacements aligned with the aforementioned views on the collected data. Our ensemble model relies on a number of similarity metrics and statistics to derive the final outcome. The paper reports on the description of the proposed model and elaborates on its validation based on various evaluation scenarios.
Soula, M., Karanika, A., Kolomvatsos, K., Anagnostopoulos, C., Stamoulis, G., 'Intelligent Tasks Allocation at the Edge based on Machine Learning and Bio-Inspired Algorithms', Evolving Systems, Springer, 2021.
Abstract

×
Current advances in the Internet of Things (IoT) and Cloud involve the presence of an additional layer between them acting as mediator for data transfer and processing in close distance to end users. This mediator is the Edge Computing (EC) infrastructure. In EC, we can identify an ecosystem of heterogeneous nodes capable of interacting with IoT devices, collecting and locally processing the data they report. The ultimate goal is to eliminate the latency we face when relying on Cloud to perform the desired processing activities. In EC, any processing is performed over a number of geo-distributed datasets formulated by the collected data that exhibit specific statistical characteristics. Processing can have the form of tasks requested by end users or applications. It becomes obvious that in the EC ecosystem, we have to carefully decide the EC nodes that will host and execute any requested task. In this paper, we extend our previous research efforts on the conclusion of efficient task allocations into the available EC nodes. We go a step forward and propose a batch processing model executed over multiple tasks and study two allocation models: a scheme based on an unsupervised machine learning technique and a bio-inspired optimization algorithm. Our models enhance the autonomous behavior of entities performing the envisioned task allocations. We provide the analytical description of the problem, our solution and the advances over the state of the art. We present and evaluate the proposed algorithms and compare them with other efforts in the domain. The pros and cons of our models are revealed through the provided extensive experimental evaluation adopting real and synthetic data.
Kolomvatsos, K., Anagnostopoulos, C., 'Landmark based Outliers Detection in Pervasive Applications', in 12th International Conference on Information and Communication Systems (ICICS 2021), 24-26 May, Valencia - Spain (Virtual), 2021.
Abstract

×
The combination of the Internet of Things and the Edge Computing gives many opportunities to support innovative applications close to end users. Numerous devices present in both infrastructures can collect data upon which various processing activities can be performed. However, the quality of the outcomes may be jeopardized by the presence of outliers. In this paper, we argue on a novel model for outliers detection by elaborating on a soft approach. Our mechanism is built upon the concepts of candidate and confirmed outliers. Any data object that deviates from the population is confirmed as an outlier only after the study of its sequence of magnitude values as new data are incorporated into our decision making model. We adopt the combination of a sliding with a landmark window model when a candidate outlier is detected to expand the sequence of data objects taken into consideration. The proposed model is fast and efficient as exposed by our experimental evaluation while a comparative assessment reveals its pros and cons.
Jodelka, O., Anagnostopoulos, C., Kolomvatsos, K., 'Adaptive Novelty Detection over Contextual Data Streams at the Edge using One-class Classification', in 12th International Conference on Information and Communication Systems (ICICS 2021), 24-26 May, Valencia - Spain (Virtual), 2021.
Abstract

×
Online novelty detection is an emerging task in Edge Computing trying to identify novel concepts in contextual data streams which should be incorporated into models executed on edge nodes. We introduce an unsupervised adaptive mechanism for online novelty detection over data streams at the network edge based on the One-class Support Vector Machine; an instance of One-class Classification paradigm. Due to adjustable periodic model retraining, our mechanism timely recognises novelties and resource-efficiently adapts to data streams. Experimental evaluation and comparative assessment showcase the effectiveness and efficiency of our mechanism over real data-streams in identifying novelty conditioned on the necessary model retraining.
Kolomvatsos, K., 'Data-driven Type-2 Fuzzy Sets for Tasks Management at the Edge', IEEE Transactions on Emerging Topics in Computational Intelligence (TETCI), 2021.
Abstract

×
Tasks allocation at the edge of the network is a significant research topic for the upcoming new era of the intelligent edge mesh. One can easily detect interesting attempts to define novel algorithms for distributing tasks into a number of heterogeneous edge nodes. Nodes interact in very dynamic environments, thus, their availability/capability of efficiently executing tasks in real time varies. In this paper, we propose a model for allocating tasks under the uncertainty present in an edge computing environment. The uncertainty is related to the current status of edge nodes and their availability for performing the requested processing activities. To manage this uncertainty, we adopt a Type-2 Fuzzy Logic system and propose a novel approach for delivering the appropriate fuzzy sets for input and output variables. Our methodology is fully adapted to nodes status as exposed by statistical reports they send at pre-defined intervals. We propose a data-driven approach that delivers the upper and lower bounds of our Type-2 fuzzy sets and present the model for concluding these bounds. We incorporate the uncertainty management mechanism into the decision making model of edge nodes being responsible to select the most appropriate peers for offloading tasks that are not possible to be executed locally. We present the performance of the proposed model through an extensive experimental evaluation and reveal its pros and cons.
Oikonomou, P., Karanika, A., Anagnostopoulos, C., Kolomvatsos, K., 'On the Use of Intelligent Models towards Meeting the Challenges of the Edge Mesh', ACM Computing Surveys, 2021.
Abstract

×
Nowadays, we are witnessing the advent of the Internet of Things (IoT) with numerous devices performing interactions between them or with their environment. The huge number of devices leads to huge volumes of data that demand the appropriate processing. The legacy approach is to rely on Cloud where increased computational resources can realize any desired processing. However, the need for supporting real time applications requires a reduced latency in the provision of outcomes. Edge Computing (EC) comes as the solver of the latency problem. Various processing activities can be performed at EC nodes having direct connection with IoT devices. A number of challenges should be met before we conclude a fully automated ecosystem where nodes can cooperate or understand their status to efficiently serve applications. In this paper, we perform a survey of the relevant research activities towards the vision of Edge Mesh (EM), i.e., a cover of intelligence upon the EC. We present the necessary hardware and discuss research outcomes in every aspect of EC/EM nodes functioning. We present technologies and theories adopted for data, tasks and resource management while discussing how machine learning and optimization can be adopted in the domain.
Kolomvatsos, K., Anagnostopoulos, C., 'Proactive, Uncertainty-Driven Queries Management at the Edge', Future Generation Computer Systems (FGCS), Elsevier, 2021.
Abstract

×
Research community has already revealed the challenges of data processing when performed at the Cloud that may affect the performance of any desired application. The main challenge is the increased latency observed when the data should `travel' to the Cloud from the location they are collected and the waiting time for getting the final response. In an Internet of Things (IoT) scenario, this time could be critical for supporting real time applications. A solution to the discussed problem is the adoption of an Edge Computing (EC) approach where data can be processed close to their collection point. IoT devices could report data to a number of edge nodes that behave as distributed data repositories having the capability of processing them and producing analytics. Analytics should match the requirements of queries defined by end users or applications with the collected data and the characteristics of every edge node. However, when a query is defined, we should identify the appropriate edge node(s) to process it. In this paper, we propose an uncertainty management model to efficiently allocate every incoming query to the available edge nodes. Our scheme adopts the principles of Fuzzy Logic (FL) theory and provides a decision making mechanism for the entity having the responsibility of the envisioned allocations. We combine the proposed uncertainty management scheme with a machine learning model based on a Support Vector Machine (SVM) to enhance the FL reasoning. Our aim is to manage all the hidden aspects of the problem combining two different technologies with different orientations. We also propose a methodology for the automated generation of the Footprint of Uncertainty (FoU) of membership functions involved in our interval Type-2 FL model. Our experimental evaluation aims at revealing the pros and cons of our mechanism presenting the results of extensive simulations adopting datasets found in the literature and a comparative analysis with other efforts in the domain.
Kolomvatsos, K., 'Proactive Tasks Management for Pervasive Computing Applications', Journal of Network and Computer Applications (JNCA), Elsevier, 2021.
Abstract

×
Current advances in the Internet of Things (IoT) and Edge Computing (EC) involve numerous devices/nodes present at both `layers' being capable of performing simple processing activities close to end users. This approach targets to limit the latency that users face when consuming the provided services. The minimization of the latency requires for novel techniques that deliver efficient schemes for tasks management at the edge infrastructure and the management of the uncertainty related to the status of edge nodes during the decision making as proposed in this paper. Tasks should be executed in the minimum time especially when we aim to support real time applications. In this paper, we propose a new model for the proactive management of tasks' allocation to provide a decision making model that results the best possible node where every task should be executed. A task can be executed either locally at the node where it is initially reported or in a peer node, if this is more efficient. We focus on the management of the uncertainty over the characteristics of peer nodes when the envisioned decisions should be realized. The proposed model aims at providing the best possible action for any incoming task. For such purposes, we adopt an unsupervised machine learning technique. We present the problem under consideration and specific formulations accompanied by the proposed solution. Our extensive experimental evaluation with synthetic and real data targets to reveal the advantages of the proposed scheme.
Ghanduri, F., Kolomvatsos, K., Anagnostopoulos, C., 'Predictive Intelligence & Explainability Challenges in Finance: A Review', In Emerging Intelligent Decision Management Systems: Applications and Challenges, Eds Prasant Kumar Pattnaik, Himansu Das, Wiley, 2021.
Abstract

×
Predictive analytics is a branch of advanced analytics that takes data, processes it and produced prediction of unknown future events. Many techniques are used as part of predictive analytics including statistics, modelling, data mining and machine learning for the analysis of current data to make these predictions. Predictive analytics in finance uses large amounts of data to determine patterns and create insights for analysts in the industry. This modern analytical solution allows AI-enabled support to human decision-making and to visualise explainability and interpretability. This chapter discusses the challenges that financial analysts face, how these challenges are approached and proposed solutions by recent researchers. These challenges include Class Imbalance, Concept Drift, Feature Selection, Nonlinearity and Time series. Explainable and interpretable solutions are also further explored, and tools are suggested as extensions of predictive analytics solutions.
Aladwani, T., Kolomvatsos, K., Deligianni, F., Anagnostopoulos, C., 'Intelligent Data Management in UAV-enabled Mobile Edge Computing: A Review', in In Emerging Intelligent Decision Management Systems: Applications and Challenges, Eds Prasant Kumar Pattnaik, Himansu Das, Wiley, 2021.
Abstract

×
The ground mobile edge computing (G-MEC) is a well-known paradigm when addressing certain real time applications in civilian as well as military areas. Several new technologies such as 5G, 6G, and smart cities have increased the demand for a new mobile edge computing (MEC) paradigm that can serve users and applications in real time. An MEC that enables unmanned vehicles has been introduced that includes MEC computing nodes such as unmanned aerial vehicles (UAVs), unmanned surface vehicles (USVs), and internet of vehicle (IoV), all of which can help implement MEC in mobility situations and make them more accessible to the end user (EU). This chapter focuses significantly on UAVs-enabled MEC as a new promising MEC paradigm that can provide the MEC architecture for numerous challenging real-life scenarios, such as desert areas, emergency response, military training, or disaster relief. However, UAVs generally have limited resources and limited energy compared with G-MEC. Hence, processing contextual data in UAVs is considered as a non-trivial challenge. This has led to artificial intelligence (AI) technology being adopted to provide several solutions to address such problems effectively. Numerous researchers have been trying to use UAVs as an ideal platform for local contextual data processing. Machine learning (ML) and deep learning (DL) paradigms (including supervised and unsupervised learning and reinforcement learning) have been applied with UAVs to resolve certain challenges effectively such as data offloading, data caching, resource allocation, and data scheduling. This chapter will highlight specific works and methods in this context and provide a comparative assessment with related work by focusing on game theory and traditional cloud-based processing methods.
Fountas, P., Kolomvatsos, K., Anagnostopoulos, C., 'A Deep Learning Model for Data Synopses Management in Pervasive Computing Applications', Computing Conference, , July 15-16, London, 2021.
Abstract

×
Pervasive computing involves the placement of processing units and services close to end users to support intelligent applications that will facilitate their activities. With the advent of the Internet of Things (IoT) and the Edge Computing (EC), one can find more room for placing services at various points in the interconnection of the aforementioned infrastructures. Of significant importance is the processing of the collected data to provide analytics and knowledge. Such processing can be realized upon the EC nodes that exhibit increased computational capabilities compared to IoT devices. An ecosystem of intelligent nodes is created at the EC giving the opportunity to support cooperative models towards the provision of the desired analytics. Nodes become the hosts of geo-distributed datasets formulated by the reports of IoT devices. Upon the datasets, a number of queries/tasks can be executed either locally or remotely. Queries/tasks can be offloaded for performance reasons to deliver the most appropriate response. However, an offloading action should be carefully designed being always aligned with the data present to the hosting node. In this paper, we present a model to support the cooperative aspect in the EC infrastructure. We argue on the delivery of data synopses distributed in the ecosystem of EC nodes making them capable to take offloading decisions fully aligned with data present at every peer. Nodes exchange their data synopses to inform their peers. We propose a scheme that detects the appropriate time to distribute the calculated synopsis trying to avoid the network overloading especially when synopses are frequently extracted due to the high rates at which IoT devices report data to EC nodes. Our approach involves a deep learning model for learning the distribution of calculated synopses and estimate future trends. Upon these trends, we are able to find the appropriate time to deliver synopses to peer nodes. We provide the description of the proposed mechanism and evaluate it based on real datasets. An extensive experimentation upon various scenarios reveals the pros and cons of the approach by giving numerical results.
Kolomvatsos, K., 'A Proactive Uncertainty Driven Model for Data Synopses Management in Pervasive Applications', 6th IEEE International Conference on Data Science and Systems (DSS), , 14-16, Fiji, 2020.
Abstract

×
Pervasive computing applications deal with the incorporation of intelligent components around end users to facilitate their activities. Such applications can be provided upon the vast infrastructures of the Internet of Things (IoT) and Edge Computing (EC). IoT devices collect ambient data transferring them towards the EC and Cloud for further processing. EC nodes could become the hosts of distributed datasets where various processing activities take place. The future of EC involves numerous nodes interacting with the IoT devices and themselves in a cooperative manner to realize the desired processing. A critical issue for concluding this cooperative approach is the exchange of data synopses to have EC nodes informed about the data present in their peers. Such knowledge will be useful for decision making related to the execution of processing activities. In this paper, we propose an uncertainty driven model for the exchange of data synopses. We argue that EC nodes should delay the exchange of synopses especially when no significant differences with historical values are present. Our mechanism adopts a Fuzzy Logic (FL) system to decide when there is a significant difference with the previous reported synopses to decide the exchange of the new one. Our scheme is capable of alleviating the network from numerous messages retrieved even for low fluctuations in synopses. We analytically describe our model and evaluate it through a large set of experiments. Our experimental evaluation targets to detect the efficiency of the approach based on the elimination of unnecessary messages while keeping immediately informed peer nodes for significant statistical changes in the distributed datasets.
Kolomvatsos, K., 'Probabilistic Data Allocation in Pervasive Computing Applications', 19th International Conference on Ubiquitous Computing and Communications (IUCC), in conjunction with the 18th IEEE International Symposium on Parallel and Distributed Processing with Applications (ISPA), December 17-19, 2020, Exeter, UK.
Abstract

×
Current advances in Pervasive Computing (PC) involve the adoption of the huge infrastructures of the Internet of Things (IoT) and the Edge Computing (EC). Both, IoT and EC, can support innovative applications around end users to facilitate their activities. Such applications are built upon the collected data and the appropriate processing demanded in the form of requests. To limit the latency, instead of relying on Cloud for data storage and processing, the research community provides a number of models for data management at the EC. Requests, usually defined in the form of tasks or queries, demand the processing of specific data. A model for pre-processing the data preparing them and detecting their statistics before requests arrive is necessary. In this paper, we propose a promising and easy to implement scheme for selecting the appropriate host of the incoming data based on a probabilistic approach. Our aim is to store similar data in the same distributed datasets to have, beforehand, knowledge on their statistics while keeping their solidity at high levels. As solidity, we consider the limited statistical deviation of data, thus, we can support the storage of highly correlated data in the same dataset. Additionally, we propose an aggregation mechanism for outliers detection applied just after the arrival of data. Outliers are transferred to Cloud for further processing. When data are accepted to be locally stored, we propose a model for selecting the appropriate datasets where they will be replicated for building a fault tolerant system. We analytically describe our model and evaluate it through extensive simulations presenting its pros and cons.
Kolomvatsos, K., Anagnostopoulos, C., 'A Deep Learning Model for Demand-driven, Proactive Tasks Management in Pervasive Computing', , IoT, MDPI, 2020.
Abstract

×
Pervasive computing applications deal with the intelligence surrounding users that can facilitate their activities. This intelligence is provided in the form of software components incorporated in embedded systems or devices in close distance with end users. One example infrastructure that can host intelligent pervasive services is the Edge Computing (EC) ecosystem. EC nodes can execute a number of tasks for data collected by devices present in the Internet of Things (IoT). In this paper, we propose an intelligent, proactive tasks management model based on the demand. Demand depicts the number of users or applications interested in using the available tasks in EC nodes, thus, characterizing their popularity. We rely on a Deep Machine Learning (DML) model and more specifically on a Long Short Term Memory (LSTM) network to learn the distribution of demand indicators for each task and estimate the future interest for them. This information is combined with historical observations and support a decision making scheme to conclude which tasks will be offloaded due to limited interest on them. We have to notice that in our decision making, we also take into consideration the load that every task may add to the processing node where it will be allocated. The description of our model is accompanied by a large set of experimental simulations for evaluating the proposed mechanism. We provide numerical results and reveal that the proposed scheme is capable of deciding on the fly while concluding the most efficient decisions.
Kolomvatsos, K., Anagnostopoulos, C., Koziri, M., Loukopoulos, T., 'Proactive & Time-Optimized Data Synopsis Management at the Edge', , IEEE Transactions on Knowledge and Data Engineering (IEEE TKDE), 2020.
Abstract

×
Internet of Things offers the infrastructure for smooth functioning of autonomous context-aware devices being connected towards the Cloud. Edge Computing (EC) relies between the IoT and Cloud providing significant advantages. One advantage is to perform local data processing (limited latency, bandwidth preservation) with real time communication among IoT devices, while multiple nodes become hosts of the collected data (reported by IoT devices). In this work, we provide a mechanism for the exchange of data synopses (summaries of extracted knowledge) among EC nodes that are necessary to give the knowledge on the data present in EC environments. The overarching aim is to intelligently decide on when nodes should exchange data synopses in light of efficient execution of tasks. We enhance such a decision with a stochastic optimization model based on the Theory of Optimal Stopping. We provide the fundamentals of our model and the relevant formulations on the optimal time to disseminate data synopses to network edge nodes. We report a comprehensive experimental evaluation and comparative assessment related to the optimality achieved by our model and the positive effects on EC.
Fountas, P., Kolomvatsos, K., 'Ensemble based Data Imputation at the Edge', , 32nd International Conference on Tools with Artificial Intelligence (ICTAI 2020), November 9-11, 2020. [Best Student Paper Award]
Abstract

×
Edge Computing (EC) offers an infrastructure that acts as the mediator between the Cloud and the Internet of Things (IoT). The goal is to reduce the latency that we enjoy when relying on Cloud. IoT devices interact with their environment to collect data relaying them towards the Cloud through the EC. Various services can be provided at the EC for the immediate management of the collected data. One significant task is the management of missing values. In this paper, we propose an ensemble based approach for data imputation that takes into consideration the spatio-temporal aspect of the collected data and the reporting devices. We propose to rely on the group of IoT devices that resemble to the device reporting missing data and enhance its data imputation process. We continuously reason on the correlation of the reported streams and efficiently combine the available data. Our aim is to `aggregate' the local view on the appropriate replacement with the `opinion' of the group. We adopt widely known similarity techniques and a statistical modelling methodology to deliver the final outcome. We provide the description of our model and evaluate it through a high number of simulations adopting various experimental scenarios.
Kolomvatsos, K., Anagnostopoulos, C., 'An Intelligent Edge-Centric Queries Allocation Scheme based on Ensemble Models', ACM Transactions on Internet Technology, 2020.
Abstract

×
The combination of Internet of Things (IoT) and Edge Computing (EC) can assist in the delivery of novel applications that will facilitate end users activities. Data collected by numerous devices present in the IoT infrastructure can be hosted into a set of EC nodes becoming the subject of processing tasks for the provision of analytics. Analytics are derived as the result of various queries defined by end users or applications. Such queries can be executed in the available EC nodes to limit the latency in the provision of responses. In this paper, we propose a meta-ensemble learning scheme that supports the decision making for the allocation of queries to the appropriate EC nodes. Our learning model decides over queries and nodes characteristics. We provide the description of a matching process between queries and nodes after concluding the contextual information for each envisioned characteristic adopted in our meta-ensemble scheme. We rely on widely known ensemble models, combine them and offer an additional processing layer to increase the performance. The aim is to result a subset of EC nodes that will host each incoming query. Apart from the description of the proposed model, we report on its evaluation and the corresponding results. Through a large set of experiments and a numerical analysis, we aim at revealing the pros and cons of the proposed scheme.
Savva, F., Anagnostopoulos, C., Triantafillou, P., Kolomvatsos, K., 'Large-scale Data Exploration using Explanatory Regression Functions', ACM Transactions on Knowledge Discovery from Data, 2020.
Abstract

×
Analysts wishing to explore multivariate data spaces, typically issue queries involving selection operators, i.e., range or equality predicates, which define data subspaces of potential interest. Then, they use aggregation functions, the results of which determine a subspace s interestingness for further exploration and deeper analysis. However, Aggregate Query (AQ) results are scalars and convey limited information and explainability about the queried subspaces for enhanced exploratory analysis. Analysts have no way of identifying how these results are derived or how they change w.r.t query (input) parameter values. We address this shortcoming by aiding analysts to explore and understand data subspaces by contributing a novel explanation mechanism based on Machine Learning. We explain AQ results using functions obtained by a three-fold joint optimization problem which assume the form of explainable piecewise-linear regression functions. A key feature of the proposed solution is that the explanation functions are estimated using past executed queries. These queries provide a coarse grained overview of the underlying aggregate function (generating the AQ results) to be learned. Explanations for future, previously unseen AQs can be computed without accessing the underlying data and can be used to further explore the queried data subspaces, without issuing more queries to the backend analytics engine.We evaluate the explanation accuracy and efficiency through theoretically grounded metrics over real-world and synthetic datasets and query workloads.
A. Karanika. P. Oikonomou, K. Kolomvatsos, C. Anagnostopoulos, 'An Ensemble Interpretable Machine Learning Scheme for Securing Data Quality at the Edge', International IFIP Cross Domain (CD) Conference for Machine Learning & Knowledge Extraction (MAKE) - CD-MAKE 2020, August 25-27, 2020.
Abstract

×
Data quality is a significant issue for any application that requests for analytics to support decision making. It becomes very important when we focus on Internet of Things (IoT) where numerous devices can interact to exchange and process data. IoT devices are connected to Edge Computing (EC) nodes to report the collected data, thus, we have to secure data quality not only at the IoT but also at the edge of the network. In this paper, we focus on the specific problem and propose the use of interpretable machine learning to deliver the features that are important to be based for any data processing activity. Our aim is to secure data quality, at least, for those features that are detected as significant in the collected datasets. We have to notice that the selected features depict the highest correlation with the remaining in every dataset, thus, they can be adopted for dimensionality reduction. We focus on multiple methodologies for having interpretability in our learning models and adopt an ensemble scheme for the final decision. Our scheme is capable of timely retrieving the final result and efficiently selecting the appropriate features. We evaluate our model through extensive simulations and present numerical results. Our aim is to reveal its performance under various experimental scenarios that we create varying a set of parameters adopted in our mechanism.
P. Fountas, K. Kolomvatsos, 'A Continuous Data Imputation Mechanism based on Streams Correlation', 10th Workshop on Management of Cloud and Smart City Systems, in conjunction with IEEE Symposium on Computers and Communications (ISCC), 2020.
Abstract

×
The increased adoption of the Internet of Things (IoT) for the delivery of intelligent applications ver huge volumes of data opens new opportunities to draw conclusions from data and support efficient decision making. For this reason many applications have been developed for data collection and processing. A large part of them are aligned with the requirements of the vast infrastructure of IoT. However, one of the biggest problems occurring at real-time applications is that they are prone to missing values. Missing values can negatively affect the outcomes of any processing activity, thus, they can limit the performance of IoT applications. In this paper, we depart from the relevant literature and propose a data imputation model that is based on the correlation of data reported by different IoT devices. Our aim is to support data imputation using the `knowledge' of the team of IoT devices over their reports for various phenomena. Our scheme adopts a continuous correlation detection methodology applied at real time reports of the involved devices. Hence, any missing value can be replaced by the aggregated outcome of data reported by correlated devices. We provide the description of our approach and evaluate it through a high number of simulations adopting various experimental scenarios.
Yiannis Kathidjiotis, Kostas Kolomvatsos, Christos Anagnostopoulos, 'Predictive Intelligence of Reliable Analytics in Distributed Computing Environments', Springer Applied Intelligence, 2020.
Abstract

×
Lack of knowledge in the underlying data distribution in distributed large-scale data can be an obstacle when issuing analytics & predictive modelling queries. Analysts find themselves having a hard time finding analytics/exploration queries that satisfy their needs. In this paper, we study how exploration query results can be predicted in order to avoid the execution of `bad'/non-informative queries that waste network, storage, financial resources and time in a distributed computing environment. The proposed methodology involves clustering of a training set of exploration queries along with the cardinality of the results (score) they retrieved and then using query-centroid representatives to proceed with predictions. After the training phase, we propose a novel refinement process to increase the reliability of predicting the score of new unseen queries based on the refined query representatives. Comprehensive experimentation with real datasets shows that more reliable predictions are acquired after the proposed refinement method, which increases the reliability of the closest centroid and improves predictability under the right circumstances.
Karanika, A., Oikonomou, P., Kolomvatsos, K., Loukopoulos, T., 'A Demand-driven, Proactive Tasks Management Model at the Edge', in IEEE International Conference on Fuzzy Systems (FUZZ-IEEE), IEEE World Congress on Computational Intelligence (WCCI), Glasgow, UK, 2020.
Abstract

×
Tasks management is a very interesting research topic for various application domains. Tasks may have the form of analytics or any other processing over the available data. The target is to efficiently allocate and execute tasks to produce meaningful results that will facilitate any decision making. The advent of the Internet of Things (IoT) and Edge Computing (EC) defines new requirements for tasks management. Such requirements are related to the dynamic environment where IoT devices and EC nodes act and process the collected data. The statistics of the collected data and the status of IoT/EC nodes are continuously updated. In this paper, we propose a demand- and uncertainty-driven tasks management scheme with the target to allocate the computational burden to the appropriate places. As the proper place, we consider the local execution of a task in an EC node or its offloading to a peer node. We provide the description of the problem and give details for its solution. The proposed mechanism models the demand for each task and efficiently selects the place where it will be executed. We adopt statistical learning and fuzzy logic to support the appropriate decision when tasks' execution is requested by EC nodes. Our experimental evaluation involves extensive simulations for a set of parameters defined in our model. We provide numerical results and reveal that the proposed scheme is capable of deciding on the fly while selecting the most efficient allocation.
K. Kolomvatsos, C. Anagnostopoulos, 'A Probabilistic Model for Assigning Queries at the Edge', in Computing, Springer, 2020.
Abstract

×
Data management at the edge of the network can increase the performance of applications as the processing is realized close to end users limiting the observed latency in the provision of responses. A typical data processing involves the execution of queries/tasks defined by users or applications asking for responses in the form of analytics. Query/task execution can be realized at the edge nodes that can undertake the responsibility of delivering the desired analytics to the interested users or applications. In this paper, we deal with the problem of allocating queries to a number of edge nodes. The aim is to support the goal of eliminating further the latency by allocating queries to nodes that exhibit a low load and high processing speed, thus, they can respond in the minimum time. Before any allocation, we propose a method for estimating the computational burden that a query/task will add to a node and, afterwards, we proceed with the final assignment. The allocation is concluded by the assistance of an ensemble similarity scheme responsible to deliver the complexity class for each query/task and a probabilistic decision making model. The proposed scheme matches the characteristics of the incoming queries and edge nodes trying to conclude the optimal allocation. We discuss our mechanism and through a large set of simulations and the adoption of benchmarking queries, we reveal the potentials of the proposed model supported by numerical results.
Ivanov, H., Anagnostopoulos, C., Kolomvatsos, K., 'In-Network Machine Learning Predictive Analytics: A Swarm Intelligence Approach', in Convergence of Artificial Intelligence and the Internet of Things, Mastorakis, G., et al., Springer, 2020.
Abstract

×
This chapter addresses the problem of collaborative Predictive Modelling via in-network processing of contextual information captured in Internet of Things (IoT) environments. In-network predictive modelling allows the computing and sensing devices to disseminate only their local predictive Machine Learning (ML) models instead of their local contextual data. The data center, which can be an Edge Gate- way or the Cloud, aggregates these local ML predictive models to predict future outcomes. Given that communication between devices in IoT environments and a centralised data center is energy consuming and communication bandwidth demanding, the local ML predictive models in our proposed in-network processing are trained using Swarm Intelligence for disseminating only their parameters within the network. We further investigate whether dissemination overhead of local ML predictive models can be reduced by sending only relevant ML models to the data center. This is achieved since each IoT node adopts the Particle Swarm Optimisation algorithm to locally train ML models and then collaboratively with their network neighbours one representative IoT node fuses the local ML models. We provide comprehensive experiments over Random and Small World network models using linear and non-linear regression ML models to demonstrate the impact on the predictive accuracy and the benefit of communication-aware in-network predictive modelling in IoT environments.
C. Anagnostopoulos, K. Kolomvatsos, 'An Intelligent, Time-Optimized Monitoring Scheme for Edge Nodes', in Journal of Network and Computer Applications, Elsevier, vol. 148, 2019.
Abstract

×
Monitoring activities over edge resources and services are essential in today's applications. Edge nodes can monitor their status and end users/applications requirements to identify their `matching' and deliver alerts when violations are present. Violations are related to any disturbance of the desired Quality of Service (QoS). QoS values depend on a number of performance metrics and can differ among applications. In this paper, we propose the use of an intelligent mechanism to be incorporated in monitoring tools adopted by edge nodes. The proposed mechanism continually observes the realizations of performance parameters that result in specic QoS values and decides when it is the right time to `fire' mitigation actions. Hence, edge nodes are capable of changing their configuration to secure the desired QOS levels as dictated by end users/applications requirements. In our work, a mitigation action could involve either upgrades in the current services/resources or o oading tasks by transferring computational load and data to peer nodes or the Cloud. We present our model and provide formulations for the solution of the problem. A high number of simulations reveal the performance of the proposed mechanism. Our experiments show that our scheme outperforms any deterministic model dened for the discussed setting as well as other efforts found in the respective literature.
K. Kolomvatsos, C. Anagnostopoulos, 'Edge-Centric Queries Stream Management based on an Ensemble Model', in 'Advances in Integration of Intelligent Methods', Springer Volume, 2019.
Abstract

×
The Internet of Things (IoT) involves numerous devices that can interact with each other or with their environment to collect and process data. The collected data streams are guided to the Cloud for further processing and the production of analytics. However, any processing in the Cloud, even if it is supported by improved computational resources, su_ers from an increased latency. The data should travel to the Cloud infrastructure as well as the provided analytics back to end users or devices. For minimizing the latency, we can perform data processing at the edge of the network, i.e., at the edge nodes. The aim is to deliver analytics and build knowledge close to end users and devices minimizing the required time for realizing responses. Edge nodes are transformed to distributed processing points where analytics queries can be served. In this paper, we deal with the problem of allocating queries, de_ned for producing knowledge, to a number of edge nodes. The aim is to further reduce the latency by allocating queries to nodes that exhibit low load (the current and the estimated), thus, they can provide the _nal response in the minimum time. However, before the allocation, we should decide the computational burden that a query will cause. The allocation is concluded by the assistance of an ensemble similarity scheme responsible to deliver the complexity class for each query. The complexity class, thus, can be matched against the current load of every edge node. We discuss our scheme and through a large set of simulations and the adoption of benchmarking queries, we reveal the potentials of the proposed model supported by numerical results.
Karanika, A., Soula, M., Anagnostopoulos, C., Kolomvatsos, K., Stamoulis, G., 'Optimized Analytics Query Allocation at the Edge of the Network', in 12th International Conference on Internet and Distributed Computing Systems, Naples, Italy, Oct. 10-12, 2019.
Abstract

×
The new era of the Internet of Things (IoT) provides the space where novel applications will play a significant role in people's daily lives through the adoption of multiple services that facilitate everyday activities. The huge volumes of data produced by numerous IoT devices make the adoption of analytics imperative to produce knowledge and support efficient decision making. In this setting, one can identify two main problems, i.e., the time required to send the data to Cloud and wait for getting the final response and the distributed nature of data collection. Edge Computing (EC) can offer the necessary basis for storing locally the collected data and provide the required analytics on top of them limiting the response time. In this paper, we envision multiple edge nodes where data are stored being the subject of analytics queries. We propose a methodology for allocating queries, defined by end users or applications, to the appropriate edge nodes in order to save time and resources in the provision of responses. By adopting our scheme, we are able to ask the execution of queries only from a sub-set of the available nodes avoiding to demand processing activities from nodes that do not `match' to the incoming queries. Our model envisions the allocation to specific epochs and manages a batch of queries at a time. We present the formulation of our problem and the proposed solution while providing results of an extensive evaluation process that reveals the pros and cons of the proposed model.
K. Kolomvatsos, P. Papadopoulou, C. Anagnostopoulos, S. Hadjiefthymiades, 'A Spatio-Temporal Data Imputation Model for Supproting Analytics at the Edge', in 18th IFIP Conference on e-Business, e-Services, and e-Society, Norway, Sept. 2019.
Abstract

×
Current applications developed for the Internet of Things (IoT) infrastructure usually involve the processing of collected data for delivering analytics and support efficient decision making models. Such models can be adopted in various application domains becoming the key technology for innovative solutions and creating new revenue streams. The basis for any processing is data analysis, usually in the form of responses in various analytics queries de_ned by end users or other applications. However, as already noted in the respective literature, data analysis results cannot be efficient when missing values are present. The research community has already proposed various missing data imputation methods paying more attention of the statistical aspect of the problem, i.e., the detection of the distribution governing the collected data. In this paper, we propose a missing values imputation method combining machine learning and a consensus scheme. We focus on the clustering of the IoT devices assuming they observe the same phenomenon and report the collected data to the edge infrastructure. Through a sliding window approach, we try to detect the IoT nodes that report similar values to an edge node and base on them to deliver the replacement value when missing data are present. We provide the description of the problem and our model together with results retrieved by an extensive set of simulations on top of real data. Our aim is to reveal the potentials of the proposed scheme and place it in the respective literature.
E. Aleksandrova, C. Anagnostopoulos, K. Kolomvatsos, 'Machine Learning Model Updates in Edge Computing: An Optimal Stopping Theory Approach', in 18th IEEE International Symposium on Parallel and Distributed Computing, June 5-7, Amsterdam, Netherlands, 2019.
Abstract

×
This work studies a sequential decision making methodology of when to update machine learning models in Edge Computing environments given underlying changes in the contextual data distribution. The proposed model focuses on updates scheduling and takes into consideration the optimal decision time for minimizing the network overhead. At the same time it preserves the prediction accuracy of models based on the principles of the Optimal Stopping Theory (OST). The paper reports on a comparative analysis between the proposed approach and other policies proposed in the respective literature while providing an evaluation of the performances using linear and support vector regression models. Our evaluation process is realized over real contextual data streams to reveal the strengths and weaknesses of the proposed strategy.
S. Sagkriotis, K. Kolomvatsos, C. Anagnostopoulos, D. Pezaros, S. Hadjiefthymiades, 'Knowledge-centric Analytics Queries Allocation in Edge Computing Environments', in IEEE Symposium on Computers and Communications (ISCC), June 29th - July 3rd, Barcelona, Spain, 2019.
Abstract

×
The Internet of Things (IoT) involves a huge number of devices that can collect data and send them to the Cloud infrastructure. The processing of data at the Cloud is characterized by increased latency in providing responses to queries defined by users or applications. Hence, the Edge Computing (EC) comes into the scene to provide a point for data processing close to end users. The collected data can be stored in edge devices and queries can be executed there to reduce the latency. In this paper, we envision a scenario where entities located in the Cloud undertake the responsibility of receiving queries and decide the edge nodes where these queries will be executed. The decision is based on the statistical information of the datasets present in nodes and the matching between the statistics and the incoming queries. Query management entities perform a grouping of the available nodes based on a set of criteria and adopt a sequential decision making. For each query, they initially select the appropriate group, excluding a number of nodes present in the remaining clusters, and, in the consecutive step, they select the (sub)set of nodes where the query will be executed. Nodes send their statistical information at pre-defined intervals to support the decision process. We describe the proposed model and analytically present each component of the scheme. Our evaluation is concluded through a high number of experimental simulations where the advantages and the shortcomings of the model are revealed.
K. Kolomvatsos, 'An Efficient Scheme for Applying Software Updates in Pervasive Computing Applications', Journal of Parallel and Distributed Computing, Elsevier, vol. 128, 2019, pp. 1-14.
Abstract

×
The Internet of Things (IoT) offers a vast infrastructure of numerous interconnected devices capable of communicating and exchanging data. Pervasive computing applications can be formulated on top of the IoT involving nodes that can interact with their environment and perform various processing tasks. Any task is part of intelligent services executed in nodes or the back end infrastructure for supporting end users' applications. In this setting, one can identify the need for applying updates in the software/firmware of the autonomous nodes.Updates are extensions or patches important for the efficient functioning of nodes. Legacy methodologies deal with centralized approaches where complex protocols are adopted to support the distribution of the updates in the entire network. In this paper, we depart from the relevant literature and propose a distributed model where each node is responsible to, independently, initiate and conclude the update process. Nodes monitor a set of metrics related to their load and the performance of the network and through a time-optimized scheme identify the appropriate time to conclude the update process. We report on an infinite horizon optimal stopping model on top of the collected performance data. The aim is to make nodes capable of identifying when their performance and the performance of the network are of high quality to efficiently conclude the update process. We provide specific formulations and the analysis of the problem while extensive simulations and a comparison assessment reveal the advantages of the proposed solution.
K. Kolomvatsos, 'A Distributed, Proactive Intelligent Scheme for Securing Quality in Large Scale Data Processing', accepted for publication in Springer Computing, 2019.
Abstract

×
The involvement of numerous devices and data sources in the current form of Web leads to the collection of vast volumes of data. The advent of the Internet of Things (IoT) enhances the devices to act autonomously and transforms them into information and knowledge producers. The vast infrastructure of the Web/IoT becomes the basis for producing data either in a structured or in an unstructured way. In this paper, we focus on a distributed scheme for securing the quality of data as collected and stored in multiple partitions. A high quality is achieved through the adoption of a model that identifies changes in the accuracy of the collected data. The proposed scheme determines if the incoming data negatively affect the accuracy of the already present datasets and when this is the case, it excludes them from further processing. We are based on a scheme that also identifies the appropriate partition where the incoming data should be allocated. We describe the proposed scheme and present simulation and comparison results that give insights on the pros and cons of our solution.
K. Kolomvatsos, C. Anagnostopoulos, 'Multi-criteria Optimal Task Allocation at the Edge', Elsevier Future Generation Computer Systems, vol. 93, 2019, pp. 358-372.
Abstract

×
In Internet of Things (IoT), numerous nodes produce huge volumes of data that are subject of various processing tasks. Tasks execution on top of the collected data can be realized either at the edge of the network or at the Fog/Cloud. Their management at the network edge may limit the required time for concluding responses and return the final outcome/analytics to end-users or applications. IoT nodes, due to their limited computational and resource capabilities, can execute a limited number of tasks over the collected contextual data. A challenging decision is related to which tasks IoT nodes should execute locally. Each node should carefully select such tasks to maximize the performance based on the current contextual information, e.g., tasks' characteristics, nodes' load and energy capacity. In this paper, we propose an intelligent decision making scheme for selecting the tasks that will be locally executed. The remaining tasks will be transferred to peer nodes in the network or the Fog/Cloud. Our focus is to limit the time required for initiating the execution of each task by introducing a two-step decision process. The first step is to decide whether a task can be executed locally; if not, the second step involves the sophisticated selection of the most appropriate peer to allocate it. When, in the entire network, no node is capable of executing the task, it is, then, sent to the Fog/Cloud facing the maximum latency. We comprehensively evaluate the proposed scheme demonstrating its applicability and optimality at the network edge.
K. Kolomvatsos, 'Time-Optimized Management of Mobile IoT Nodes for Pervasive Applications', Journal of Network and Computer Applications, Elsevier, vol. 125, 2019, pp. 155-167.
Abstract

×
The Internet of Things (IoT) incorporates numerous nodes adopted to support novel pervasive computing applications. Nodes are capable of interacting each other and/or collect/process huge volumes of ambient data. Any service or application executed on top of the collected data is hosted by the operating software/firmware of nodes, thus, such software should be up-to-date. Legacy techniques dealing with the update task cannot efficiently support it due to the adopted centralized approach that suffers from a number of disadvantages. In this paper, we go a step forward and propose a time-optimized and network performance-aware model for initiating and concluding the update process. Our aim is to have the nodes independently deciding the initiation of the update process by finding the appropriate time to execute it. Every node acts autonomously and monitors the network's performance to find a slot where performance parameters advocate for an efficient and uninterrupted conclusion of the update task. Hence, the proposed model can be adapted to the environment and the status of each node. The final decision is made taking into consideration multiple parameters and it is based on the solution of the widely known Secretary Problem (SP) originated in the Optimal Stopping Theory (OST). We provide the description of the problem, specific formulations and the analysis of our solution while extensive experiments reveal the advantages of the proposed scheme.
K. Kolomvatsos, 'An Intelligent, Uncertainty Driven Management Scheme for Software Updates in Pervasive IoT Applications', Elsevier Future Generation Computer Systems, vol. 83, pp. 116-131, 2018.
Abstract

×
The era of the Internet of Things (IoT) involves a huge number of autonomous devices (nodes) capable of monitoring and interacting with their environment. The autonomous devices are also capable of being interconnected, thus, they are able to exchange data. Pervasive computing applications can be built on top of this infrastructure offering efficient solutions for multiple domains. Nodes can execute intelligent, light-weight processing of the collected data being capable of responding in case of events. Apart from the software necessary to perform the discussed processing tasks, nodes are coming with pre-installed software necessary to perform basic functionalities e.g., communication. When nodes act in dynamic environments, it is necessary to update both, the software necessary for their functioning and the software necessary to perform processing tasks and respond to potential events. Updates are necessary as they involve software extensions and patches important for the efficient functioning of the IoT nodes. In this paper, we propose a distributed updates management model enhancing the autonomous nature of the IoT nodes. Legacy models deal with centralized approaches (i.e., a central server) where complex algorithms are adopted to derive the models / protocols for the distribution of the updates. In our approach, each node is responsible to, independently, initiate and conclude the update process, thus, keeping the installed software up-to-date. The central server is responsible only for indicating when the updates are offered to the nodes. Every node monitors a set of performance metrics (either for the node itself or the network) and based on an intelligent scheme decides the appropriate time to conclude the update process. We adopt an ensemble forecasting model on top of a pool of estimators and an optimization model to derive the right time for initiating and concluding the update process. We are based on the solution of the known Santa Fe bar problem to perform load balancing in the network. The aim is to have the nodes deciding the conclusion of the update process in different time intervals, thus, to keep the load of the network to low levels. We provide specific formulations and the analysis of our problem while extensive simulations and a comparison assessment reveal the advantages of the proposed solution.
K. Kolomvatsos, 'Time-Optimized Management of IoT Nodes', Elsevier Ad Hoc Networks, vol. 69, 2018, pp. 1-14.
Abstract

×
The vision of the Internet of Things (IoT) aims to offer a vast infrastructure of numerous interconnected devices. IoT provides the necessary techniques for defining and identifying the intelligent nodes. IoT consists of the basis for pervasive computing technologies that aim to provide solutions where numerous intelligent devices can interact with their environment and perform simple processing tasks. Intelligent applications can be built on top of this infrastructure leading to significant advances in various domains. As numerous devices interact in very dynamic environments, the need for new services is imperative. One can identify also the need for applying updates in the software / firmware of the autonomous nodes (i.e., the intelligent devices). Such updates are necessary as they include software extensions and patches very important for the smooth functioning of the IoT nodes. Legacy methodologies involve centralized approaches where complex algorithms and protocols are adopted to derive the means for the distribution of the updates to the nodes. In this paper, we propose a distributed approach where each node is responsible to initiate and conclude the update process. Nodes decide independently when they will conclude the update process. We envision that each node monitors specific performance metrics (either for the node itself or the network) and based on a time-optimized scheme identifies the appropriate time to perform the update process. We propose the adoption of a finite horizon optimal stopping scheme on top of the monitored metrics. Our stopping model originates in the Optimal Stopping Theory} (OST) and takes into consideration multiple performance metrics at the same time. The aim is to have the nodes capable of identifying when their performance and the performance of the network are of high quality in order to conclude the update process without disturbing them from the execution of the assigned tasks. We provide specific formulations and the analysis of our problem while extensive experiments and a comparison assessment reveal the advantages of the proposed solution.
K. Kolomvatsos, C. Anagnostopoulos, 'Intelligent Applications over Large-Scale Data Streams', The Scottish Informatics & Computer Science Alliance (SICSA), DemoFest, Edimburgh, Scotland, Nov. 6th, 2018.
Abstract

×
In this poster, we present our research titled INNOVATE. INNOVATE aims at a novel holistic solution for orchestrating the efficient analytics tasks allocation and execution of numerous exploratory and inferential modelling queries over large-scale data streams. INNOVATE focuses on bio-mimetic & bio-inspired management of ensembles of statistical learners and ecosystems of streaming data Query Controllers & Processors. INNOVATE will establish the framework to develop intelligent applications for providing inferential analytics and in-network predictive modelling in (near) real time. Real-time insights will be the basis for timely, high quality services that will increase end-users' and data analysts' satisfaction & improve quality of exploratory analysis.
K. Kolomvatsos, C. Anagnostopoulos, 'In-Network Edge Intelligence for Optimal Task Allocation', 30th International Conference on Tools with Artificial Intelligence, Nov. 5-7, Volos, Greece, 2018.
Abstract

×
The presence of numerous nodes in the Internet of Things (IoT) provides a framework where multiple services and applications can be efficiently executed. Huge volumes of data are produced by IoT devices offering the basis for delivering intelligent inferential and predictive analytics. Every IoT node may execute a number of tasks on top of the collected and sensed data, thus, supporting novel task-oriented applications. Processing data at the network edge, i.e., at the nodes, significantly reduces the time for deriving analytics and increases the performance. However, the computational capabilities of nodes are limited and they can only support a short number of tasks. Demanding tasks like inferential analytics are crucial for supporting local real-time applications, however, they may deplete nodes' resources. In this paper, we propose an optimal distributed in-network processing model that pushes the intelligence of task allocation decision at the network edge. Our methodology derives the scheduling of alternative task execution when a node decides that a task could not be executed locally. Nodes sophisticatedly determine their networked peers where the task can be assigned to. The decision is autonomously taken in each node and when no appropriate peer node is considered for the execution of a task, the task is then delegated to the Fog/Cloud for possible processing. Local optimal task allocation decision is made on multiple contextual parameters involving load, bandwidth, computational power, and data size in the edge nodes. We comprehensively evaluate the proposed scheme through simulations demonstrating its applicability and optimality at the network edge.
K. Kolomvatsos, C. Anagnostopoulos, 'An Edge-Centric Ensemble Scheme for Queries Assignment', in 8th International Workshop on Combinations of Intelligent Methods and Applications in conjunction with the 30th International Conference on Tools with Artificial Intelligence, Nov. 5-7, Volos, Greece, 2018.
Abstract

×
The new era of the Internet of Things (IoT) reveals new potentials for the management of numerous devices. Such devices produce data streams that are guided to the Cloud for further processing. However, any processing in the Cloud, even if it is supported by increased computational resources, suffers from increased latency. For minimizing the latency we can perform data processing at the edge of the network, i.e., at the edge nodes. The aim is to provide analytics and build knowledge on top of the collected data in the minimum time. In this paper, we deal with the problem of allocating queries, defined for producing knowledge, to a number of edge nodes. The aim is to further reduce the latency by allocating queries to nodes that exhibit low load (the current and the estimated), thus, they can provide the final response in the minimum time. However, before the allocation, we should decide the computational burden that a query will add. The allocation is concluded by the assistance of an ensemble similarity scheme responsible to deliver the complexity class for each query. The complexity class, thus, can be matched against the current load of every edge node. We discuss our scheme and through a large set of simulations and the adoption of benchmarking queries, we reveal the potentials of the proposed model supported by numerical results.
K. Kolomvatsos, T. Loukopoulos, 'Scheduling the Execution of Tasks at the Edge', 2018 IEEE Conference on Evolving and Adaptive Intelligent Systems (EAIS 2018), Rhodes, Greece, May 25-27, 2018.
Abstract

×
Numerous devices present in the Internet of Things (IoT) infrastructure are capable of collecting data from their environment and conclude simple processing tasks. These devices are also able to store an amount of data necessary to perform the envisioned tasks and produce knowledge. Such knowledge can be adopted to take decisions related to the presence of events. Every device, i.e., an IoT node, should keep the necessary data locally to reduce the latency in providing responses instead of relying to Cloud. Nodes act as a team and cooperate to store the collected data close to the processing of tasks that are in the form of queries coming from nodes themselves or end users. In this paper, we propose a model for deciding the allocation of data in a set of nodes. Initially, every node decides if the incoming data are correlated with the available stored datasets or they are outliers. We propose an ensemble scheme for multidimensional outliers detection that results, in real time, the final decision. When data are accepted to be locally stored, nodes select their peers where data will be replicated. This way, we keep the data in multiple locations in the network aiming to reduce latency in the provision of responses and support a fault tolerant mechanism. The replication decision is based on the correlation of the incoming data with the present datasets. We analytically describe our model and evaluate it through extensive simulations presenting its pros and cons.
K. Kolomvatsos, M. Koziri, T. Loukopoulos, 'An Intelligent Scheme for the Identification of QoS Violations in Virtualized Environments', 30th International Conference on Tools with Artificial Intelligence, Nov. 5-7, Volos, Greece, 2018.
Abstract

×
Current networking applications involve the definition and utilization of multiple virtualized resources on top of the available infrastructure. Software defined networks increase the performance compared with legacy systems as any functionality is managed through software. Securing the quality of service in such environments is significant to support novel applications that deliver their results in real time. In this paper, we propose a monitoring mechanism that observes the performance of the virtualized resources and identifies possible quality of service violations. Our model can be applied to any application domain, however, it is adapted to virtualized resources. We rely on a simple model that collects performance data, focuses on multiple parts of a virtualized functions chain and immediately concludes potential violations in real time. The proposed mechanism is incorporated in an SDN controller that is responsible to manage the virtualized resources. We provide an analytical description of the model and through a large set of simulations, we reveal its performance. Our results exhibit the timely identification of quality of service violations even in very dynamic environments where the performance of the network changes continuously.
K. Kolomvatsos, 'An Intelligent Scheme for Assigning Queries', Springer Applied Intelligence, https://doi.org/10.1007/s10489-017-1099-5, 2018.
Abstract

×
Analytics provided on top of large scale data streams are the key research subject for future decision making applications. The huge volumes of data make their partitioning imperative to efficiently support novel applications. Such applications should be based on intelligent, efficient methods for querying multiple data partitions. A processor is placed in front of each partition dedicated to manage/execute queries for the specific piece of data. Continuous queries over these data sources require intelligent mechanisms to result the final outcome (query response) in the minimum time with the maximum performance. This paper proposes a mechanism for handling the behaviour of an entity that undertakes the responsibility of handling the incoming queries. Our mechanism adopts a time-optimized scheme for selecting the appropriate processor(s) for each incoming query through the use of the Odds algorithm. We try to result the optimal assignment, i.e., queries to processors, in the minimum time while maximizing the performance. We provide mathematical formulations for describing the discussed problem and present simulation results and a comparative analysis. Through a large number of experiments, we reveal the advantages of the model and give numerical results comparing it with a deterministic model as well as with other efforts in the domain.
K. Kolomvatsos, P. Oikonomou, M. Koziri, T. Loukopoulos, 'A Distributed Data Allocation Scheme for Autonomous Nodes', IEEE International Conference on Scalable Computing and Communications, Guangzhou, China, Oct. 8-12, 2018.
Abstract

×
Numerous devices present in the Internet of Things (IoT) infrastructure are capable of collecting data from their environment and conclude simple processing tasks. These devices are also able to store an amount of data necessary to perform the envisioned tasks and produce knowledge. Such knowledge can be adopted to take decisions related to the presence of events. Every device, i.e., an IoT node, should keep the necessary data locally to reduce the latency in providing responses instead of relying to Cloud. Nodes act as a team and cooperate to store the collected data close to the processing of tasks that are in the form of queries coming from nodes themselves or end users. In this paper, we propose a model for deciding the allocation of data in a set of nodes. Initially, every node decides if the incoming data are correlated with the available stored datasets or they are outliers. We propose an ensemble scheme for multidimensional outliers detection that results, in real time, the final decision. When data are accepted to be locally stored, nodes select their peers where data will be replicated. This way, we keep the data in multiple locations in the network aiming to reduce latency in the provision of responses and support a fault tolerant mechanism. The replication decision is based on the correlation of the incoming data with the present datasets. We analytically describe our model and evaluate it through extensive simulations presenting its pros and cons.
K. Kolomvatsos, P. Papadopoulou, S. Hadjiefthymiades, 'Data Storage in Internet of Things: A Proposed Distributed model', 12th Mediterranean Conference on Information Systems, Corfu, Greece, Sept. 28-30, 2018.
Abstract

×
In Internet of Things, numerous devices are able to collect and report data while they can execute simple processing tasks to produce knowledge. Nodes exhibit limited computational resources, thus, they can only perform a limited number of tasks and store a short version of the collected data. In this paper, we propose a scheme that focuses on a distributed model for data storage in a group of nodes. Nodes cooperate with each other exchanging statistical information for their data. Our work aims to provide a model for the selection of the node where the incoming data should be stored irrelevantly of the node in which they are initially reported. The selection process involves a recommendation system that is based on the statistical similarity of the incoming data with the datasets present in the group, the load of each node and the in-network communication cost. Our distributed model acts proactively and tries to store similar data to the same nodes. The aim is to have a view on the statistics of the available datasets beforehand, thus, facilitating the post-processing and the production of knowledge. The proposed recommendation system adopts statistical measures, forecasting techniques and a mul-ticlass classification model to deliver the final result. We report on the evaluation of our scheme and present experimental results towards the presentation of pros and cons of our model.
K. Kolomvatsos, C. Anagnostopoulos, A. Marnerides, Q. Ni, S. Hadjiefthymiades, D. Pezaros, ‘Uncertainty-driven Ensemble Forecasting of QoS in Software Defined Networks’, accepted in 22nd IEEE Symposium on Computers and Communications (ISCC), Heraklion, Greece, 2017.
Abstract

×
Software Defined Networking (SDN) is the key technology for combining networking and Cloud solutions to provide novel applications. SDN offers a number of advantages as the existing resources can be virtualized and orchestrated to provide new services to the end users. Such a technology should be accompanied by powerful mechanisms that ensure the end-to-end quality of service at high levels, thus, enabling support for complex applications that satisfy end users needs. In this paper, we propose an intelligent mechanism that agglomerates the benefits of SDNs with real-time `Big Data' forecasting analytics. The proposed mechanism, as part of the SDN controller, supports predictive intelligence by monitoring a set of network performance parameters, forecasting their future values, and deriving indications on potential service quality violations. By treating the performance measurements as time-series, our mechanism employs a novel ensemble forecasting methodology to estimate their future values. Such predictions are fed to a Type-2 Fuzzy Logic system to deliver, in real-time, decisions related to service quality violations. Such decisions proactively assist the SDN controller for providing the best possible orchestration of the virtualized resources. We evaluate the proposed mechanism w.r.t. precision and recall metrics over synthetic data.
K. Kolomvatsos, S. Hadjiefthymiades, ‘Predictive Intelligence in Analytics Aggregation of Partially Ordered Subsets’,IEEE Transactions on Systems, Man and Cybernetics: Systems, 2017.
Abstract

×
The increased amount of users devices lead to a huge amount of data that should be efficiently managed by modern applications. Data are reported through streams and, usually, are stored into a number of partitions. Data separation techniques offer a lot of advantages as applications can process them in parallel. Progressive analytics are adopted to deliver partial responses and, thus, to save time in applications execution. Of great importance are data exploration and analytics queries that report on the ordered set of objects according to a predefined function. Such queries require novel intelligent predictive mechanisms for deriving responses based on the partial results retrieved by the distributed data partitions. We assume a set of finite query processors placed in front of each partition of data. Such processors are responsible to deliver progressive analytics to a Query Controller (QC). These progressive analytics are related to partial ordered subsets of objects on top of each partition of data. The QC receives queries, assigns the analytics task to the underlying processors and optimally decides on the right time to deliver the analytics results to the application. The proposed QC applies specific time-optimized techniques and aggregation operators for deriving the final response, i.e., ordered set of objects, over streams of partial responses, i.e., partial ordered subsets. The final decision is affected by variations in the aggregated set. We perform a comprehensive performance assessment of our mechanism with synthetic streaming data and report on the throughput of the QC, the quality of the final ordered set of objects and the latency for delivering the final result.
K. Kolomvatsos, S. Hadjiefthymiades, ‘Learning the Engagement of Query Processors for Intelligent Analytics’, Springer Applied Intelligence Journal, vol. 46(1), 96-112, pp. 1-17, 2017.
Abstract

×
Current applications require the processing of huge amounts of data produced by applications or end users personal devices. In such settings, intelligent analytics on top of large scale data are the key research subject for future data driven decision making. Due to the huge amount of data, analytics should be based on an efficient technique for querying big data partitions. Each partition contains only a part of the data and a processor is dedicated to execute queries for the corresponding partition. A Query Controller (QC) is responsible for managing continuous queries and returning the final outcome to users / applications by using the underlying processors. In this paper, we propose a learning scheme to be adopted by the QC for allocating each query to the available processors. We adopt the Q-learning algorithm to calculate the reward that the QC obtains for every allocation between queries and processors. The outcome is an efficient model that derives the optimal allocation for the incoming queries. We provide mathematical formulations for solving the discussed problem and present our simulation results. Through a large number of simulations, we reveal the advantages of the proposed model and give numerical results while comparing our framework with a baseline model.
K. Kolomvatsos, ‘An Intelligent, Uncertainty Driven Aggregation Scheme for Streams of Ordered Sets’, Springer Applied Intelligence (APIN), doi 10.1007/s10489-016-0789-8, pp. 1-23, 2016.
Abstract

×
Data streams management has attracted the attention of many researchers as numerous devices generate huge amounts of data. Data are reported through streams and stored into a number of partitions. Separation techniques facilitate the parallel management of data, however, intelligent methods are necessary to manage these multiple instances. Progressive analytics could be adopted to deliver partial responses and, possibly, to save time in execution of applications. An important research domain is the efficient management of queries over multiple partitions. Usually, such queries demand responses in the form of ordered sets of objects (e.g., top-k queries). These sets include objects in a ranking order and require novel mechanisms for deriving responses based on partial results. In this paper, we study a setting of multiple partitions and propose an intelligent, uncertainty driven decision making mechanism that aims to respond to streams of queries. The mechanism delivers an ordered set of objects over a number of partial ordered subsets retrieved by each partition. A number of query processors are placed in front of each partition and report progressive analytics to a Query Controller (QC). The QC receives queries, assigns the task to the underlying processors and decides the right time to deliver the final ordered set of objects to the application. We propose an aggregation model for deriving the final set of objects and a Fuzzy Logic (FL) inference process. We present a Type-2 FL system that decides when the QC should stop aggregating partial subsets and return the final response to the application. We report on the performance of the proposed mechanism through the execution of a large set of experiments. Our results deal with the throughput of the QC, the quality of the final ordered set of objects and the time required for delivering the final response.
K. Kolomvatsos, C. Anagnostopoulos and S. Hadjiefthymiades, ‘A Time Optimized Scheme for Top-k List Maintenance over Incomplete Data Streams’, Elsevier Information Sciences (INS), vol. 311, pp. 59-73, 2015.
Abstract

×
Incomplete data streams are the main subject for a large number of research efforts. Such efforts, usually, focus on the creation and maintenance of top-k lists useful to provide effective responses to top-k queries. In case of large amount of data received in high rates the problem becomes more intense as an efficient method for maintaining the top-k list is judged imperative. In this paper, we focus on the behavior of an Observer Entity (OE) responsible to observe the incoming data and initiate the maintenance process of the top-k list. The maintenance process involves the calculation of the scores for each object and the update of the top-k list. We adopt the principles of Optimal Stopping Theory (OST) and discuss a scheme that results in the appropriate time where the OE decides to initiate the maintenance process. Due to the incomplete data setting, the high rate of the incoming data and the large amount of data, the maintenance process is initiated only when the OE has the appropriate amount of information to provide an effective top-k list. In contrast to other research efforts, we do not keep any additional sets of objects. Instead, we attempt to minimize the number of the necessary processes over the list of objects. We present a mathematical analysis of our scheme and an extensive experimental evaluation. The comparison with other models shows that the proposed model provides efficiency in the list management and minimizes the required time to result the final top-k list.
K. Kolomvatsos, C. Anagnostopoulos, S. Hadjiefthymiades, ‘An Efficient Time-Optimized Scheme for Progressive Analytics in Big Data’, Elsevier Big Data Research, vol. 2(4), 2015, pp. 155-165.
Abstract

×
Big data analytics is the key research subject for future data driven decision making applications. Due to the large amount of data, progressive analytics could provide an efficient way for querying big data clusters. Each cluster contains only a piece of the examined data. Continuous queries over these data sources require intelligent mechanism to result the final outcome (query response) in the minimum time with the maximum performance. A Query Controller (QC) is responsible to manage continuous queries and return the final outcome to users or applications. In this paper, we propose a mechanism to be adopted by the query controller. The proposed mechanism could manage partial results retrieved by a number of processors each one responsible for each cluster. Each processor executes a query over the specific cluster of data. The proposed mechanism adopts two models for handling the incoming partial results. The first is based on a finite horizon model and the second is based on an infinite horizon model. We provide mathematical formulations for solving the discussed problem and present simulation results. Through a large number of experiments, we reveal the advantages of the proposed models and give numerical results comparing them with a deterministic model. These results indicate that the proposed models can efficiently reduce the required time for returning the final outcome to the user / application while keeping the quality of the aggregated result at high levels.

Contextual Reasoning

C. Anagnostopoulos, K. Kolomvatsos, 'Predictive Intelligence to the Edge through Approximate Collaborative Context Reasoning', Springer Applied Intelligence, vol. 48(4), 2018, pp. 966-991.
Abstract

×
We focus on an Internet of Things (IoT) environment where a network of sensing and computing devices are responsible to locally process contextual data, reason and collaboratively infer the appearance of a specific phenomenon (event). Pushing processing and inference to the edge of the IoT network allows the complexity of the event reasoning process to be distributed into many smaller and more manageable pieces and to be physically located at the source of the contextual information it needs to work on. This enables a huge amount of rich contextual data to be processed in real time that would be prohibitively complex and costly to deliver on a traditional centralized cloud/back-end processing system. We propose a distributed adaptive, multiple-context perspective event reasoning model under uncertainty on each IoT device (e.g., environmental sensor). Each device senses and processes context data and reasons on the presence of an event based on different local context perspectives: (i) expert knowledge on event representation, (ii) outliers inference, and (iii) deviation from locally predicted context. Such kind of approximate reasoning is achieved through a contextualized, Type-2 Fuzzy belief-driven clustering process, where clusters of devices are formed according to their belief on the presence of an event. Our distributed intelligence model efficiently identifies any localized abnormality on the contextual data in light of event reasoning through aggregating local degrees of belief, updates, and adjusts its knowledge to contextual data outliers and novelty detections. We provide comprehensive experimental evaluation and comparison assessment of the proposed distributed model over real contextual data and show the benefits stemmed from its adoption in an IoT environment.
K. Kolomvatsos, C. Anagnostopoulos, S. Hadjiefthymiades, ‘Distributed Localized Contextual Event Reasoning under Uncertainty’, IEEE Internet of Things Journal, vol. 4(1), 2017, pp. 183-191.
Abstract

×
We focus on Internet of Things (IoT) environments where sensing and computing devices (nodes) are responsible to observe, reason, report and react to a specific phenomenon. Each node captures context from data streams and reasons on the presence of an event. We propose a distributed predictive analytics scheme for localized context reasoning under uncertainty. Such reasoning is achieved through a contextualized, knowledge-driven clustering process, where the clusters of nodes are formed according to their belief on the presence of the phenomenon. Each cluster enhances its localized opinion about the presence of an event through consensus realized under the principles of Fuzzy Logic (FL). The proposed FL-driven consensus process is further enhanced with semantics adopting Type-2 Fuzzy Sets to handle the uncertainty related to the identification of an event. We provide a comprehensive experimental evaluation and comparison assessment with other schemes over real data and report on the benefits stemmed from its adoption in IoT environments.
K. Kolomvatsos, C. Anagnostopoulos, S. Hadjiefthymiades, ‘Data Fusion & Type-2 Fuzzy Inference in Contextual Data Stream Monitoring’, IEEE Transactions on Systems, Man and Cybernetics: Systems, vol. PP, Issue 99, pp.1-15, 2016.
Abstract

×
Data stream monitoring mechanisms have a wide acceptance in many application domains. The reason is that such mechanism provides the basis for building intelligent applications that undertake the responsibility of deriving decisions based on data streams retrieved by the environment. For instance, in environmental monitoring, a central system receives sensors measurements and derives decisions related to any abnormalities. Abnormalities are related to negative effects that, probably, heavily affect human lives. A number of sensors can be spread in a specific area and undertake the responsibility of monitoring environmental parameters for specific phenomena (e.g., fire, flood). In this paper, we propose a mechanism that builds on top of sensors streaming contextual data and proceeds with the appropriate decisions upon immediate identification of phenomena. The proposed mechanism involves data fusion, consensus theory and time-series prediction for efficient sensors measurements aggregation. In addition, the mechanism adopts Type-2 Fuzzy Sets theory for handling the uncertainty in the phenomena identification phase. We perform a set of simulations over real data and report on the advantages and disadvantages of the proposed mechanism. Finally, we compare our mechanism with a Type-1 Fuzzy Logic system to demonstrate their performance in terms of false alarms in phenomena identification under uncertain real measurements.
C. Anagnostopoulos, K. Kolomvatsos, ‘A Delay-Resilient and Quality-aware Mechanism over Incomplete Contextual Data Streams’, Elsevier Information Sciences Journal (INS), vol. 355-356, pp. 90-109, 2016.
Abstract

×
We study the case of scheduling a Contextual Information Process (CIP) over incomplete multivariate contextual data streams coming from sensing devices in Internet of Things (IoT) environments. CIPs like data fusion, concept drift detection, and predictive analytics adopt window-based methods for processing continuous stream queries. CIPs involve the continuous evaluation of functions over contextual attributes (e.g., air pollutants measurements from environmental sensors) possibly incomplete (i.e., containing missing values) thus degrading the quality of the CIP results. We introduce a mechanism, which monitors the quality of the contextual streaming values and then optimally determines the appropriate time to activate a CIP. CIP is optimally delayed in hopes of observing in the near future higher quality of contextual values in terms of validity, freshness and presence. Our time-optimized mechanism activates a CIP when the expected quality is maximized taking also into account the induced cost of delay and an aging framework of freshness over contextual values. We propose two analytical time-based stochastic optimization models and provide extensive sensitivity analysis. We provide a comparative assessment with sliding window-centric models found in the literature and showcase the efficiency of our mechanism on improving the quality of results of a CIP.
C. Anagnostopoulos, S. Hadjiefthymiades, K. Kolomvatsos, ‘Accurate, Dynamic & Distributed Localization of Phenomena for Mobile Sensor Networks’, ACM Transactions on Sensor Networks (TOSN), vol. 12(2), art. No 9, 2016.
Abstract

×
We present a robust, dynamic scheme for the automatic self-deployment and relocation of mobile sensor nodes around areas where phenomena take place. Our scheme aims (a) to sense environmental parameters and accurately capture the evolution of a certain phenomenon (e.g., fire, air contamination) and (b) to fully automate the deployment process by letting nodes relocate, self-organize (and self-reorganize) and optimally cover the focus area. Our intention is to `opportunistically' modify the previous (possibly uniform) sensor placement to attain better spatio-temporal phenomena monitoring. The needed intelligence is fully distributed within the mobile sensor network so that the deployment algorithm is executed incrementally by different nodes. The presented algorithm adopts the Particle Swarm Optimization technique which yields very promising results in the undertaken performance assessment. Our findings show that the presented algorithm captures a certain phenomenon with very high accuracy while maintaining the global energy expenditure of the network at low levels. Random occurrences of similar phenomena put stress upon the algorithm which manages to react promptly and efficiently manage the available sensing resources in the broader setting.
K. Kolomvatsos, C. Anagnostopoulos, S. Hadjiefthymiades, ‘Contextual Reasoning under Uncertainty in Sensor Data Stream Monitoring’, International Journal of Monitoring and Surveillance Technologies Research, vol. 3(2), pp. 1-19, 2015.
Abstract

×
Data streams monitoring plays an important role in the identification of abnormalities in many application domains. Abnormalities are related to negative effects that, consequently, affect people's quality of living. A number of sensors could be placed in various locations undertaking the responsibility of monitoring specific phenomena. Sensors report their measurements to a central system that is capable of situational reasoning. The system, through decision making, responds to any event related to the observed phenomena. In this paper, we propose a mechanism that builds on top of the sensors measurements and derives the appropriate decisions for the immediate identification of events. The proposed system adopts data fusion and prediction (time series regression) methods for efficiently aggregating sensors measurements. We also adopt Fuzzy Logic for handling the uncertainty on the decision making. We perform a set of simulations over real data and report on the advantages and disadvantages of the proposed system.
C. Anagnostopoulos, S. Hadjiefthymiades and K. Kolomvatsos, ‘Time Optimized User Grouping in Location Based Services’, Elsevier Computer Networks (COMNET), vol. 81, pp. 220-244, 2015.
Abstract

×
We focus on Location Based Services (LBSs) which deliver information to groups of mobile users based on their spatial context. The existence (and non-trivial lifetime) of groups of mobile users can simplify the operation of LBS and reduce network overhead (due to location updates and application content flow). We propose an incremental group formation algorithm and an optimally scheduled, adaptive group validation mechanism which detects and further exploits spontaneous formation of groups of mobile users. The advantages of application simplification and network load reduction are pursued. Through the proposed mechanism the LBS server (back-end system) monitors representatives of formed groups, i.e., group leader, and, in turn, disseminates location-dependent application content to group members. We first introduce an incremental group formation algorithm for group partitioning and identification of group leaders. We elaborate on the mechanism which adopts the Optimal Stopping Theory in order to assess the group persistence through evaluation of compactness and coherency metrics. We compare the performance of the proposed scheme with that of existing mechanisms for moving objects clustering and quantify the benefits stemming for its adoption.
K. Kolomvatsos, C. Anagnostopoulos, S. Hadjiefthymiades, ‘Intelligent Contextual Data Stream Monitoring’, 8th International Conference on Pervasive Technologies Related to Assistive Environments, July 2015, Corfu, Greece, 2015.
Abstract

×
Contextual data monitoring plays an important role in increasing the quality of life of humans. Sensors observing specific activities report contextual data to a central system capable of situational reasoning. The system responds to any event related to the observed phenomenon. We propose an intelligent mechanism that builds on top of sensors measurements and derives the appropriate decisions for immediate identification of events. The mechanism adopts multivariate data fusion, time-series prediction, and consensus theory for aggregating measurements. We adopt Fuzzy Logic for handling the induced uncertainty in the decision making on the derived alerts. Simulations over real contextual data showcase the advantages and disadvantages of our monitoring mechanism.
K. Kolomvatsos, C. Anagnostopoulos, S. Hadjiefthymiades, ‘An Efficient Environmental Monitoring System adopting Data Fusion, Prediction and Fuzzy Logic’, 6th International Conference on Information, Intelligence, Systems and Applications, Corfu, Greece, July 2015.
Abstract

×
Environmental monitoring plays an important role in the identification of abnormalities in the environment's characteristics. Abnormalities are related to negative effects that, consequently, heavily affect human lives. A number of sensors could be placed in a specific area and undertake the responsibility of monitoring environment's characteristics for specific phenomena. Sensors report back their measurements to a central system capable of situational reasoning. Accordingly, the system through decision making responds to any event related to the observed phenomenon. In this paper, we propose a mechanism that builds on top of the sensors measurements and derives the appropriate decisions for the immediate identification of events. The proposed system adopts data fusion and prediction (time series regression) statistical learning methods for efficiently aggregating sensors measurements. We also adopt Fuzzy Logic for handling the uncertainty on the decision making on the derived alerts. We perform a set of simulations over real data and report on the advantages and disadvantages of the proposed system.
C. Anagnostopoulos, K. Kolomvatsos and S. Hadjiefthymiades, 'Efficient Location Based Services for Groups of Mobile Users', in IEEE 14th International Conference on Mobile Data Management (MDM), Milan, Italy, June 3-6, 2013.
Abstract

×
We study the performance improvement of Location Based Services through the identification and subsequent use of groups of mobile nodes. In our scheme we exploit the formation of nodes into groups in order to reduce the computation load incurred in back end systems (e.g., Location Servers) and the associated network overhead. The back end systems track the position and communicate with the Group Leader (GL). The GL, in turn, passes the received information to the members of the group (e.g., through short range communications). The formation of mobile groups is validated over time to avoid misinterpreted temporary groupings which could endanger the adoption of the reduced load/overhead scheme. A time scheduling scheme based on the Optimal Stopping Theory assists in the finalization of the group validity. Metrics like group compactness are thoroughly assessed in line with the optimal stopping time scheme to increase confidence on group validity and persistence. Performance assessment reveals significant benefits for the considered location based services system.
V. Papataxiarhis, V. Riga, V. Nomikos, O. Sekkas, K. Kolomvatsos, V. Tsetsos, P. Papageorgas, S. Vourakis, S. Hadjiefthymiades, and G. Kouroupetroglou, 'MNISIKLIS: Indoor Location Based Services for All', In Proc. of the 5th International Symposium on LBS and TeleCartography, Nov. 26th - 28th, Salzburg, Austria, 2008.
Abstract

×
MNISIKLIS is an integrated system aiming to provide universal, indoor location-based services focusing on navigation. This paper presents the overall MNISIKLIS architecture and certain implementation details. In the context of the Design for All approach, the system targets to the support of several types of users, including persons with disabilities as well as elderly, by exploiting multimodal interaction. Moreover, the system implements efficient path finding algorithms and provides advanced user experience through highly personalized services. MNISIKLIS adopts Semantic Web technologies (e.g., ontologies and reasoning methods) for representing and managing application models. Furthermore, MNISIKLIS exploits modern positioning techniques in order to achieve high quality positioning. The paper discusses the algorithms and the models that accommodate the services provided by the system. Additionally, an analysis of the positioning subsystem, the user interaction subsystem and the peripheral infrastructure is given. Hence, a new paradigm in the area of location-based systems is presented.

Fuzzy Logic Reasoning and Applications

Kolomvatsos, K., Kalouda, M., Papadopoulou, P., Hadjiefthymiades, S., 'Fuzzy Trust Modeling for Pervasive Computing Applications', Journal of Data Intelligence, 2021.
Abstract

×
Pervasive computing applications involve the interaction between autonomous entities for performing complex tasks and producing knowledge. Autonomous entities can interact to exchange data and knowledge to fulfil applications requirements. Intelligent Agents (IAs) activated in various devices offer a lot of advantages when representing such entities due to their autonomous nature that enables them to perform the desired tasks in a distributed way. However, in such open and dynamic environments, IAs should be based on an efficient mechanism for trusting unknown entities when exchanging data. The trust level of an entity should be automatically calculated based on an efficient methodology. Each entity is uncertain for the characteristics and the intentions of the others. Fuzzy Logic (FL) seems to be the appropriate tool for handling such kind of uncertainty. In this paper, we present a model for trust calculation under the principles of FL. Our scheme takes into consideration the social dimension of trust as well as personal experiences of entities before they decide interactions with an IA. The proposed model is a two-level system involving three FL sub-systems to calculate (a) the social trust (based on experiences retrieved by the community), (b) the individual trust (based on personal experiences) and (c) the final trust. We present our results by evaluating the proposed system compared to other models and reveal its significance.
Oikonomou, P., Kolomvatsos, K., Tziritas, N., Theodoropoulos, G., Loukopoulos, T., Stamoulis, G., 'Uncertainty Driven Workflow Scheduling Using Unreliable Cloud Resources', in IEEE International Symposium on Network Computing and Applications (NCA), November 24-27, 2020.
Abstract

×
The Cloud infrastructure offers to end users a broad set of heterogenous computational resources using the pay-as-you-go model. These virtualized resources can be provisioned using different pricing models like the unreliable model where resources are provided at a fraction of the cost but with no guarantee for an uninterrupted processing. However, the enormous gamut of opportunities comes with a great caveat as resource management and scheduling decisions are increasingly complicated. Moreover, the presented uncertainty in optimally selecting resources has also a negatively impact on the quality of solutions delivered by scheduling algorithms. In this paper, we present a dynamic scheduling algorithm (i.e., the Uncertainty-Driven Scheduling - UDS algorithm) for the management of scientific workflows in Cloud. Our model minimizes both the makespan and the monetary cost by dynamically selecting reliable or unreliable virtualized resources. For covering the uncertainty in decision making, we adopt a Fuzzy Logic Controller (FLC) to derive the pricing model of the resources that will host every task. We evaluate the performance of the proposed algorithm using real workflow applications being tested under the assumption of different probabilities regarding the revocation of unreliable resources. Numerical results depict the performance of the proposed approach and a comparative assessment reveals the position of the paper in the relevant literature.
K. Kolomvatsos, M. Kalouda, P. Papadopoulou, S. Hadjiefthymiades, 'A Fuzzy Trust Model for Autonomous Entities Acting in Pervasive Computing', 17th International Conference on Trust, Privacy and Security in Digital Business, Sept. 14-17, Bratislava, Slovakia, 2020.
Abstract

×
Pervasive computing applications, usually, involve the interaction between autonomous entities for performing complex tasks and producing knowledge. Autonomous entities can interact each other to exchange data and knowledge, thus, to fulfill applications' requirements. Intelligent Agents (IAs) 'activated' in various devices offer a lot of advantages when representing such entities due to their autonomous nature that gives them the capability of performing the desired tasks in a distributed manner. However, in such open and very dynamic environments, IAs should be based on an efficient mechanism for trusting unknown entities when exchanging data. The trust level of an entity should be automatically calculated based on an efficient methodology. Each entity is uncertain for the characteristics and the intentions of the others. Fuzzy Logic (FL) is the appropriate tool for handling such kind of uncertainty. In this paper, we present a model for trust calculation with the use of FL. Our scheme takes into consideration the social dimension of trust as well as personal experiences of entities before they decide interactions with n IA. The proposed model is a two-level system involving three FL sub-systems: the sub-system for calculating the social trust (based on experiences retrieved by the community), the sub-system for calculating the individual trust (based on personal experiences) and the final trust calculation sub-system. We present our results by evaluating the proposed system with end users and we show its significance.
K. Kolomvatsos, ‘Effective Problem Solving through Fuzzy Logic Knowledge Bases Aggregation’, Springer Soft Computing Journal, 20(3), 1071-1092, 2016.
Abstract

×
Cooperative problem solving has attracted great interest in the research community. A number of nodes can cooperate in order to reach a common goal which is the solution to a specific problem. Through a team effort, nodes try to achieve the best possible result. Each node undertakes the responsibility to fulfill a specific task (part of a complex plan towards the solution of the problem) and return the final outcome to a coordinator. The coordinator assigns tasks to nodes and, accordingly, collects the results. In this paper, we propose the adoption of Fuzzy Logic (FL) in the decision making process of each node. Nodes adopt the provided (by the coordinator) FL knowledge base that indicates the correct actions during the problem solving process. This knowledge base is updated during the execution of the assigned task in order to be fully aligned with the environment characteristics. Partial knowledge bases experienced by the nodes are sent back to the coordinator and are aggregated in order to generate a `global' knowledge base that incorporates the experience retrieved by the team. We describe the aggregation process and propose two models: the first indicates the immediate distribution of the aggregated FL rule base to the active nodes and the second indicates the future use of the aggregated FL rule base. Our evaluation involves the realization of the proposed framework in a specific research domain as well as numerical results retrieved by a large number of simulations. Our results show that there is a trade-off between the two proposed models concerning the quality of the final solution and the time required to retrieve the final outcome.
K. Kolomvatsos, D. Trivizakis, S. Hadjiefthymiades, ‘An Adaptive Fuzzy Logic System for Automated Negotiations’, accepted for publication in Elsevier Fuzzy Sets and Systems (FSS), 2014.
Abstract

×
The rapid growth of the Web means that humans are increasingly incapable of searching among millions of resources to find and purchase items. Autonomous entities such as agents could help in these situations. Electronic markets (EMs) are virtual sites where these autonomous entities can interact to exchange items and obtain specific returns. In this study, we consider the interactions between buyers and sellers in these EMs, where we focus specifically on the buyer side. These interactions can be modeled as finite horizon negotiations. However, the buyer cannot be certain of the characteristics of the seller during negotiations (incomplete knowledge). Thus, to address this uncertainty, we propose a fuzzy logic (FL) system that is responsible for determining the appropriate actions of the buyer during every negotiation round. We also propose an adaptation technique that updates the FL rule base and system membership functions as necessary. Using this approach, the system can respond to even the complex strategies followed by a seller. A seller strategy estimation method is also included in the system, which employs the known kernel density estimator. We provide results for a large number of negotiations and compare our system with previous research in this area. Our results show that the proposed system exhibits good performance in many negotiation scenarios.
K. Kolomvatsos, C. Anagnostopoulos, S. Hadjiefthymiades, 'Sellers in Marketplaces: A Fuzzy Logic based Decision Support System', Elsevier Information Sciences (INS) (IF 2013: 3.893), vol. 278, pp. 267-284, 2014.
Abstract

×
Web business models typically rely on environments where entities, not known in advance, try to negotiate and agree upon the exchange of products. Such environments are termed as Electronic Markets (EMs). In EMs there are two main groups of entities: the buyers and the sellers. Intelligent agents can play the role of buyers and sellers as delegates. Agents, acting autonomously, can guarantee the efficiency in items retrieval. The interaction between buyers and sellers can be modeled as a zero knowledge negotiation. In this paper, we discuss the basic characteristics of the negotiation and define a decision support mechanism for sellers. We focus on bilateral single issue negotiations between a buyer and a seller. The proposed decision making mechanism is based on Fuzzy Logic (FL) in order to handle uncertainty in the negotiation process. The seller, at every negotiation round, receives the buyer's offer and decides her line of actions. In this setting, we consider that no knowledge on the entities strategies is available. The seller uses fuzzy inference rules in order to decide if she is going to accept or reject the offer of the buyer at every negotiation round. Compared with other relevant schemes our approach exhibits efficiency by increasing the utility that the seller gains through negotiations.
Kostas Kolomvatsos and Stathes Hadjiefthymiades, 'Buyer Behaviour Adaptation Based on a Fuzzy Logic Controller and Prediction Techniques', Elsevier Fuzzy Sets and Systems (FSS) (IF 2012: 1.749), vol. 189(1), Feb. 2012, pp. 30-52.
Abstract

×
Current form of Web provides numerous product resources available to users. Users can rely on intelligent agents for purchase actions. These actions are taken in specific environments such as Electronic Markets (EMs). In this paper, we study the interaction process between buyers and sellers and focus on the buyer side. Each buyer has the opportunity to interact with a number of sellers trying to buy the most appropriate products. This interaction can be modeled as a finite horizon Bargaining Game (BG). In this game, players have opposite goals concerning the product price. We adapt a number of techniques in the buyer side trying to give the appropriate level of efficiency in the buyer decision process. The buyer uses a prediction mechanism in combination with the usage of Fuzzy Logic (FL) theory in order to be able to predict the upcoming seller proposal and thus, understand the seller pricing policy. Based on this, he/she can adapt his/her behavior when trying to purchase products. The buyer adaptation mechanism produces the belief that the buyer has about the seller pricing policy and a parameter that indicates his/her own pricing policy which yields the buyer offers in the upcoming rounds. Moreover, the buyer is based on FL system that derives the appropriate actions at every round of the BG. Our results show that the combination of Fuzzy Logic (FL) with the above mentioned techniques provides an efficient decision mechanism in the buyer side that in specific scenarios outperforms a theoretical optimal model.
K. Kolomvatsos, C. Anagnostopoulos, and S. Hadjiefthymiades, 'A Fuzzy Logic System for Bargaining in Information Markets', ACM Transactions on Intelligent Systems and Technology (ACM TIST), vol. 3(2), Feb. 2012, art. No 32.
Abstract

×
Future web business models involve virtual environments where entities interact in order to sell or buy information goods. Such environments are known as Information Markets (IMs). Intelligent agents are used in IMs for representing buyers or information providers (sellers). We focus on the decisions taken by the buyer in purchase negotiation process with sellers. We propose a reasoning mechanism on the offers (prices of information goods) issued by sellers based on Fuzzy Logic. The buyer's knowledge on the negotiation process is modeled through Fuzzy Sets. We propose a fuzzy inference engine dealing with the decisions that the buyer takes on each stage of the negotiation process. The outcome of the proposed reasoning method indicates whether the buyer should accept or reject the sellers' offers. Our findings are very promising for the efficiency of automated transactions undertaken by intelligent agents.
R. Arapoglou, K. Kolomvatsos, and S. Hadjiefthymiades, 'Buyer Agent Decision Process Based on Automatic Fuzzy Rules Generation Methods', In Proc. of the 2010 IEEE World Congress on Computational Intelligence (WCCI 2010), FUZ-IEEE, July 18th - 23rd, Barcelona, Spain, pp. 856-863.
Abstract

×
Software Agents can assume the responsibility of finding and negotiating products on behalf of their owners in an electronic marketplace. In such cases, Fuzzy Logic can provide an efficient reasoning mechanism especially for the buyer side. Agents representing buyers can rely on a fuzzy rule base in order to reason for their next action at every round of the interaction process with sellers. In this paper, we describe a model where the buyer builds its fuzzy knowledge base using algorithms for automatic fuzzy rules generation based on data provided by experts and compare a set of such algorithms. Owing to such algorithms, agent developers spend less time and effort for the definition of the underlying rule base. Moreover, the rule base is efficiently created through the use of the dataset indicating the behaviour of the buyer and, thus, representing its line of actions in the electronic marketplace. In our work, we use such algorithms for the definition of the buyer behaviour and we provide critical insides for every algorithm describing their advantages and disadvantages. Moreover, we present numerical results for every basic parameter of the interaction process, such as the time required for the rule base generation, the Joint Utility of the interaction process or the value of the acceptance degree that each algorithm results.
Kostas Kolomvatsos and Stathes Hadjiefthymiades, 'Automatic Fuzzy Rules Generation for the Deadline Calculation of a Seller Agent', In Proc. of the 9th International Symposium on Autonomous Decentralized Systems (ISADS 2009), Athens, Greece, March 23-25, 2009, pp. 429-434.
Abstract

×
Intelligent Agents can help users in finding and retrieving goods from electronic marketplaces. Additionally, agents can represent providers in such places facilitating the automatic negotiation about the purchase of products. In this paper, we describe a finite horizon bargaining model between buyers and sellers and we focus on the seller's side. Seller agents are a good example of an autonomous decentralized system. We present a method for the 'bargaining' deadline calculation based on Fuzzy-Logic (FL). Such deadline indicates the time for which it is profitable for a seller to participate in the bargaining procedure. We provide methods for automatic fuzzy rules generation. These rules result the deadline values at each interaction and are based on data provided by experts. We compare results taken from a Fuzzy controller based on such automatic methods with results taken by previous research efforts.
K. Kolomvatsos, C. Anagnostopoulos, and S. Hadjiefthymiades, 'On The Use of Fuzzy Logic in a Seller Bargaining Game', In Proc. of the 32nd Annual IEEE International Computer Software an Applications Conference (COMPSAC 2008), July 28th - August 1st, Turku, Finland, 2008, pp. 184-191.
Abstract

×
Information marketplaces are places where users search and retrieve information goods. Intelligent Agents could represent the participating entities in such places, i.e, assume the role of buyers and sellers of information products. In this paper, we introduce a finite horizon bargaining model between buyers and sellers. We examine the seller's side and define a method for the 'bargaining' deadline calculation based on Fuzzy-Logic (FL). Such deadline indicates the time for which it is profitable for a seller to participate in the bargaining procedure, i.e., the time threshold for his offers. We represent the seller's knowledge / policy adopting the Fuzzy Set Theory and provide a fuzzy inference engine for reasoning about the bargaining deadline. The result of the reasoning process defines the degree of patience of the seller agent, thus, affecting the time for which that seller participates in the bargaining game.
Kostas Kolomvatsos and Stathes Hadjiefthymiades, 'On the Use of Fuzzy Logic in Electronic Marketplaces', in the book 'Cross Disciplinary Applications of Artificial Intelligence and Pattern Recognition: Advancing Technologies', ed. Vijay Mago, IGI Global, 2011.
Abstract

×
Today, there is a large number of product providers in the Web. Electronic Marketplaces (EMs) enable entities to negotiate and trade products. Usually, intelligent agents assume the responsibility of representing buyers or sellers in EMs. However, uncertainty about the characteristics and intentions of the negotiating entities is present in these scenarios. Fuzzy Logic (FL) theory presents a lot of advantages when used in environments where entities have limited or no knowledge about their peers. Hence, entities can rely on a FL knowledge base that determines the appropriate action on every possible state. FL can be used in offers, trust, or time constraints definition or when an agent should decide during the negotiation process. The autonomic nature of agents in combination with FL leads to more efficient systems. In this chapter, we provide a critical review on the adoption of FL in marketplace systems and present our proposal for the buyer side. Moreover, we describe techniques for building FL systems focusing on clustering techniques. Our aim is to show the importance of FL adoption in such settings.

Recommender Systems

K. Kolomvatsos, M. Koziri, T. Loukopoulos, 'A Recommendation System for Allocating Video Resources in Multiple Partitions', to be published in Big Data Recommender Systems: Recent Trends and Advances, Institution of Engineering and Technology (IET), 2018.
Abstract

×
A recommendation system or recommender aims to deliver meaningful recommendations for items or services to any interested party (e.g., users, applications). Recommenders provide their results on top of the collected data related either to the items' and users' description or ratings defined by users. Recommenders can be adopted in the domain of large scale data management with significant advantages. Due to the huge volumes of data, many techniques consider the separation of data into a number of partitions. Analytics are delivered on top of these data partitions and, accordingly, are aggregated to form the final response into the incoming queries. Data separation techniques can be incorporated to allocate the data into the appropriate partitions, thus, to improve the efficiency in the delivery of analytics. In this paper, we propose a recommendation system responsible to allocate the data to the most appropriate partition according to their current contents. Our approach facilitates the provision of the analytics for each data partition by collecting 'similar' data into the same partition. The aim is to support statistical insights into every partition to efficiently define query execution plans. We adopt a decision making scheme combined with a Naive Bayesian classifier for deriving the appropriate partition. We focus on the management of streams of video files. The proposed recommender derives the appropriate partition for each incoming video file based on a set of characteristics. We evaluate our scheme through a set of simulations that reveal its strengths and weaknesses.
K. Kolomvatsos, C. Anagnostopoulos and S. Hadjiefthymiades, ‘An Efficient Recommendation System based on the Optimal Stopping Theory’, Elsevier Expert Systems with Applications (ESWA), (IF 2013: 1.965), vol. 41(15), 2014.
Abstract

×
A Recommendation System (RS) aims to deliver meaningful recommendations to users for items (e.g., music and books), which are of high interest to them. We consider an RS which directly communicates with a set of providers in order to access the information of the items (e.g., description), rate them according to the user's preferences, and deliver an item list (IL). The RS is enhanced with a mechanism, which sequentially observes the rating information (e.g., similarity degree) of the items and decides when to deliver the IL to the user, without exhausting the entire set of providers. Hence, the RS saves time and resources. We propose two mechanisms based on the theory of optimal stopping. Both mechanisms deliver an IL, which sufficiently matches to the user's needs having examined a partial set of items. That is, the number of items in the delivered IL is optimal, producing a high level of user satisfaction, i.e., Quality of Recommendation (QoR). Our simulations reveal the efficiency of the mechanisms and quantify the benefits stemming from their adoption.
B. Lika, K. Kolomvatsos, S. Hadjiefthymiades, ‘Facing the Cold Start Problem in Recommender Systems’, Elsevier Expert Systems with Applications (ESWA) (IF 2013: 1.965), vol. 41(4), 2014, 2065-2073.
Abstract

×
A recommendation system (RS) aims to deliver meaningful recommendations to users for specific items (e.g., music, books). Popular techniques involve content-based models and collaborative filtering approaches. In this paper, we deal with a very important problem in RSs: The cold start problem. This problem is related to recommendations for novel users who have not exhibited any preferences into the system or for new items. We propose a model where widely known classification algorithms in combination with similarity techniques and prediction mechanisms provide the necessary means for retrieving recommendations. The proposed system results personalized recommendations for new users using their demographic characteristics. Such characteristics help the system to identify other users with similar behavior. Our experiments show the performance of the proposed system through a large number of experiments. We adopt the widely known dataset provided by the GroupLens research group. By utilizing different users number, we reveal the advantages of the proposed solution and give numerical results.

Smart Cities Applications

K. Kolomvatsos, C. Anagnostopoulos, 'Reinforcement Machine Learning for Predictive Analytics in Smart Cities', Informatics, MDPI, 4, 16, 2017.
Abstract

×
The digitization of our lives cause a shift in the data production as well as in the required data management. Numerous nodes are capable of producing huge volumes of data in our everyday activities. Sensors, personal smart devices as well as the Internet of Things (IoT) paradigm lead to a vast infrastructure that covers all the aspects of activities in modern societies. In the most of the cases, the critical issue for public authorities (usually, local like municipalities) is the efficient management of data towards the support of novel services. The reason is that analytics provided on top of the collected data could help in the delivery of new applications that will facilitate citizens' lives. However, the provision of analytics demands intelligent techniques for the underlying data management. The most known technique is the separation of huge volumes of data into a number of parts and their parallel management to limit the required time for the delivery of analytics. Afterwards, analytics requests in the form of queries could be realized and derive the necessary knowledge for supporting intelligent applications. In this paper, we define the concept of a Query Controller (QC) that receives queries for analytics and assigns each of them to a processor placed in front of each data partition. We discuss an intelligent process for query assignments that adopts Machine Learning (ML). We adopt two learning schemes, i.e., Reinforcement Learning (RL) and clustering. We report on the comparison of the two schemes and elaborate on their combination. Our aim is to provide an efficient framework to support the decision making of the QC that should swiftly select the appropriate processor for each query. We provide mathematical formulations for the discussed problem and present simulation results. Through a comprehensive experimental evaluation, we reveal the advantages of the proposed models and describe the outcomes results while comparing them with a deterministic framework.
Anagnostopoulos, T., Kolomvatsos, K., Medvedev, A., Amirian, P., Morley, J., Hadjieftymiades, S., Zaslavsky, A., ‘Challenges and Opportunities of Waste Management in IoT-enabled Smart Cities: A Survey’, IEEE Transactions on Sustainable Computing, Volume: 2, Issue: 3, 2017.
Abstract

×
The new era of the Internet of Things (IoT) paradigm is being enabled by the proliferation of various devices like RFIDs, sensors, and actuators. IoT attaches IPs to devices having significant computational capabilities, transforming them to 'smart things'. These 'smart things' can play significant roles in the definition of novel applications to facilitate people's lives. At the same time, smart devices are embedded in the environment to monitor and collect ambient information. In a city, this leads to Smart City frameworks. Intelligent services could be offered on top of such information related to any aspect of humans' activities. A typical example of services offered in the framework of Smart Cities is IoT-enabled waste management. Waste management involves not only the collection of the waste in the field but also the transport and disposal to the appropriate locations. In this paper, we present a comprehensive and thorough survey of ICT-enabled waste management models. Specifically, we focus on the adoption of IoT as a key enabling technology in contemporary waste management. We report on the strengths and weaknesses of various models in order to reveal their characteristics. Finally, we propose a novel, mobile and IoT-enabled waste management scheme which incorporates the management of a heterogeneous fleet of waste collection vehicles. The proposed model leads to the efficient management of the available collection vehicles and, thus, to lower operational costs.
T. Anagnostopoulos, K. Kolomvatsos, C. Anagnostopoulos, A. Zaslavsky, S. Hadjiefthymiades, ‘Assessing Dynamic Models for High Priority Waste Collection in Smart Cities’, Elsevier Journal of Systems and Software (JSS), vol. 110, 2015, pp. 178-192.
Abstract

×
Waste Management (WM) represents an important part of Smart Cities (SCs) with significant impact on modern societies. WM involves a set of processes ranging from waste collection to the recycling of the collected materials. The proliferation of sensors and actuators enable the new era of Internet of Things (IoT) that can be adopted in SCs and help in WM. Novel approaches that involve dynamic routing models combined with the IoT capabilities could provide solutions that outperform existing models. In this paper, we focus on a SC where a number of collection bins are located in different areas with sensors attached to them. We study a dynamic waste collection architecture, which is based on data retrieved by sensors. We pay special attention to the possibility of immediate WM service in high priority areas, e.g., schools or hospitals where, possibly, the presence of dangerous waste or the negative effects on human quality of living impose the need for immediate collection. This is very crucial when we focus on sensitive groups of citizens like pupils, elderly or people living close to areas where dangerous waste is rejected. We propose novel algorithms aiming at providing efficient and scalable solutions to the dynamic waste collection problem through the management of the trade-off between the immediate collection and its cost. We describe how the proposed system effectively responds to the demand as realized by sensor observations and alerts originated in high priority areas. Our aim is to minimize the time required for serving high priority areas while keeping the average expected performance at high level. Comprehensive simulations on top of the data retrieved by a SC validate the proposed algorithms on both quantitative and qualitative criteria which are adopted to analyze their strengths and weaknesses. We claim that, local authorities could choose the model that best matches their needs and resources of each city.

Autonomous Behaviour and Decision Making

Parambath, S. A., Alfahad, S. A., Anagnostopoulos, C., Kolomvatsos, K., 'Sequential Block Elimination for Dynamic Pricing', in 32024 IEEE International Conference on Data Mining Workshops (ICDMW), Abu Dhabi, 2024.
Abstract

×
In this paper, we propose a Thompson Sampling based sequential block elimination approach for dynamic pricing within pure-exploration Multi-Armed Bandit (MAB) setting. Given an l-dimensional action space, which represents various price ranges, our objective is to identify the optimal prices for different alternatives or variants of a product in order to maximize the total revenue. Our contribution lies in the development of a novel block elimination-based MAB algorithm. The proposed algorithm begins by discretizing the continuous action space into a finite set of discrete actions. Subsequently, a recursive block elimination procedure is employed to progressively remove sub-optimal actions from the set. The elimination process leverages the calculation of confidence bounds over the blocks of actions, enabling the efficient exclusion of sub-optimal choices. We conduct extensive experiments on a dynamic pricing problem in the logistics domain to evaluate the effectiveness of our proposed approach. The experimental results demonstrate that our method is able to identify the optimal arms within the given budget.
Zioviris, G., Kolomvatsos, K., Stamoulis, G., 'An Intelligent Sequential Fraud Detection Model based on Deep Learning', The Journal of Supercomputing, Springer, 2024.
Abstract

×
Fraud detection and prevention has received a lot of attention from the research community due to its high impact on financial institutions’ revenues and reputation. The increased use of the Web and the provision of online services open up the pathway for exposing these systems to numerous threats and jeopardise their effective functioning. Naturally, financial frauds are increased in number and forms imposing various requirements for their efficient and immediate detection. These requirements are related to the performance of the adopted models as well as the timely response of the decision making mechanism. Machine Learning (ML) and data mining are two research domains that can provide a number of techniques/algorithms for fraud detection and setup the road for mitigation actions. However, these methods still need to be improved with respect to the detection of unknown fraud patterns and the incorporation of big data processing mechanisms. This paper presents our attempt to build a hybrid system, i.e., a sequential scheme for combining two deep learning models and efficiently detect potential financial frauds. We elaborate on the combination of an autoencoder and a Long Short-Term Memory (LSTM) recurrent neural network trained upon datasets which are processed through the use of an oversampling technique. Oversampling is adopted to handle heavily imbalanced datasets which is the ‘natural’ scenario due to the limited number of frauds compared to the humongous volumes of transactions. The proposed approach tends to capture much more fraud events in comparison with other conventional ML techniques. Our experimental evaluation exposes that our model exhibits a good performance in terms of recall & precision.
Albanis, G., Zioulis, N., Thermos, S., Chatzitofis, A., Kolomvatsos, K., 'Noise-in, Bias-out: Balanced and Real-Time MoCap Solving', in Computer Vision for Metaverse Workshop, International Conference on Computer Vision (ICCV), Paris, France, October 2-6, 2023.
Abstract

×
Panagidi, K., Riazuelo, L., Alonso, I., Murillo, A. C., Montano, L., Cantero, M., Martins, R., Kolomvatsos, K., Hadjiefthymiades, S., 'On the Innovative Management of Remote Robotic Experimentation', in the 5th Iberian Robotics Conference (ROBOT 22), Zaragoza, Spain, November 23-25, 2022.
Abstract

×
This paper presents one of the first future Internet experimentation platforms, for managing multiple UxVs in vehicular (road), aerial and maritime environments. The aim of this work is to create a federation of different network testbeds that work together to make their resources available under a common framework. We introduce a consolidated architecture, based on a multi-tier design pattern, paying special attention on the resource management and the creation of experiments over multiple testbeds interconnected. One of the main advantages of this framework is that it is agnostic to the type and OS of the platforms under each testbed. A description of the current testbeds integrated in the RAWFIE federated testbed is shown. We also present a real use case for validating the proposed architecture in one of the available testbeds. Finally, this work details how to achieve the integration of the testbed contributed means within the RAWFIE infrastructure.
Tragoudaras, A., Stoikos, P., Fanaras, K., Tziouvaras, A., Floros, G., Dimitriou, G., Kolomvatsos, K., Stamoulis, G., 'Design Space Exploration of a Sparse MobileNetV2 Using High-Level Synthesis and Sparse Matrix Techniques on FPGAs', Sensors, 22, 4813, 2022, https://doi.org/10.3390/s22124318.
Abstract

×
Convolution Neural Networks (CNNs) are gaining ground in deep learning and Artificial Intelligence (AI) domains, and they can benefit from rapid prototyping in order to produce efficient and low-power hardware designs. The inference process of a Deep Neural Network (DNN) is considered a computationally intensive process that requires hardware accelerators to operate in realworld scenarios due to the low latency requirements of real-time applications. As a result, High-Level Synthesis (HLS) tools are gaining popularity since they provide attractive ways to reduce design time complexity directly in register transfer level (RTL). In this paper, we implement a MobileNetV2 model using a state-of-the-art HLS tool in order to conduct a design space exploration and to provide insights on complex hardware designs which are tailored for DNN inference. Our goal is to combine design methodologies with sparsification techniques to produce hardware accelerators that achieve comparable error metrics within the same order of magnitude with the corresponding state-of-the-art systems while also significantly reducing the inference latency and resource utilization. Toward this end, we apply sparse matrix techniques on a MobileNetV2 model for efficient data representation, and we evaluate our designs in two different weight pruning approaches. Experimental results are evaluated with respect to the CIFAR-10 data set using several different design methodologies in order to fully explore their effects on the performance of the model under examination.
Zioviris, G., Kolomvatsos, K., Stamoulis, G., 'Credit Card Fraud Detection using a Deep Learning Multistage Model', Journal of Supercomputing, 2022.
Abstract

×
The banking sector is on the eve of a serious transformation, and the thrust behind it is Articial Intelligence (AI). Innovative AI applications have already been proposed to deal with problems in areas like credit scoring, risk assessment, client experience, and portfolio management. Recently, deep learning models have been introduced to detect and forecast possible fraud transactions with augmented eciency being compared to the conventional machine learning methods and statistics. Such methods gain signicant popularity due to their ability to estimate the unknown distribution of the collected data, thus, increasing their capability of detecting more complex fraud events. In this paper, we introduce a novel multistage deep learning model that combines a feature selection process built upon autoencoder models (Simple & Variational) and a deep Convolutional Neural Network (CNN) to detect frauds. To manage highly unbalanced datasets, we adopt and test several oversampling techniques. We describe the problem under consideration and our contribution that provides the solution for it. An extensive set of experimental scenarios reveals the performance of the proposed scheme as exposed by the relevant numerical results. Finally, a comparative assessment is adopted to prove the superiority of our model, under certain conditions, compared with a SVM and typical CNN.
Albanis, G., Chatzitofis, A., Thermos, S., Zioulis, N., Kolomvatsos, K., 'Towards Scalable and Real-Time Markerless Motion Capture', in IEEE Conference on Virtual Reality, 12-16 March, 2022.
Abstract

×
Human motion capture and perception without the need for complex systems with specialized cameras or wearable equipment is the holy grail for many human-centric applications. Here, we present a scalable markerless motion capture method that estimates 3D human poses in real-time using low-cost hardware. We do so by replacing the inefficient 3D joint reconstruction techniques, such as learnable triangulation and feature splatting, with a novel uncertainty-driven approach that exploits the available depth information and the edge sensors spatial alignment to fuse the per viewpoint estimates into final 3D joint positions.
Panagiotis Fountas, Maria Papathanasaki, Kostas Kolomvatsos, Nikos Tziritas, 'Proactive Attack Detection at the Edge through an Ensemble Deep Learning Model', in , 20th International Conference on Ubiquitous Computing and Communications (IUCC), 20-22 December, London, UK, 2021.
Abstract

×
The new form of the Web involves numerous devices present in two infrastructures, i.e., the Internet of Things (IoT) and the Edge Computing (EC) infrastructure. IoT devices are adopted to record ambient data and host lightweight processing to provide support for applications offered to end users. EC is placed between the IoT and Cloud and can be the host of more advanced processing activities. It has gained popularity due to the increased computational resources compared to the IoT and the decreased latency in the provision of responses compared to the Cloud. A high number of nodes may be present at the EC that should secure the Quality of Service (QoS) of the desired applications. Apparently, EC nodes become the central point of the process where the collected data are circulated and processed in the network. This approach imposes various security issues that should be mitigated in order to maintain high QoS levels and the uninterrupted functioning of EC nodes. In this paper, motivated by the need of the increased security, we propose an ensemble scheme for the detection of attacks in the EC. Our distributed scheme relies on the adoption of deep learning algorithms to enhance the intelligence of nodes and proactively detect potential malfunctions. Our model is embedded in EC nodes and is continuously applied upon the streams of data transferred by IoT devices in an upwards mode to the Cloud. We present the details of the proposed approach and evaluate it through a variety of simulation scenarios. Our intention is to reveal the strengths and weaknesses of the provided model when adopt in a very dynamic environment like the EC.
Panagiotis Oikonomou, Nikos Tziritas, Kostas Kolomvatsos, Thanasis Loukopoulos, 'Fast Heuristics for Mixed Fleet Capacitated Multiple TSP with Applications to Location Based Games and Drone Assisted Transportation', in 25th Pan-Hellenic Conference on Informatics,November 26-28, Volos, Greece, 2021.
Abstract

×
The Travelling Salesman Problem (TSP) is one of the most known problems in combinatorics and consists of finding a route so that a salesman visits all destination points exactly once with minimum cost. Applications of the problem have long been studied in various fields of computer science, however, with the modernization of transportation means as manifested by autonomous vehicles and drone based delivery, variations of TSP have seen renewed interest. Simultaneously, in location based games such as Pokemon GO, the problem of defining optimal routes in order to direct a single player or a team of players into visiting a set of locations, leads to solving TSP variations. Inspired by the above, in this paper we tackle the problem of Mixed Fleet Capacitated Multiple TSP (mfcmTSP), whereby a set of couriers (using different vehicle types or drones) must deliver a set of homogeneous parcels to predefined destinations with minimum makespan. All transportation means have capacity measured as the number of parcels they can carry. We motivate mfcmTSP considering a scenario whereby all parcels are available at a single base station and a fleet of one truck with unlimited capacity, a motorbike with k capacity and a drone with capacity of one parcel are available to fulfill delivery orders. We present a fast heuristic to obtain good solution quality with negligible running time overhead and compare it against a yardstick strategy whereby only the truck is used. Experiments demonstrate the merits of our approach.
Nikos Pappas, Panagiotis Oikonomou, Nikos Tziritas, Kostas Kolomvatsos, Thanasis Loukopoulos, 'Bin Packing Heuristics for the Multiple Workflow Scheduling Problem', in 25th Pan-Hellenic Conference on Informatics,November 26-28, Volos, Greece, 2021.
Abstract

×
In the multiple workflow scheduling problem a set of workflows has to be scheduled concurrently onto system s available resources. Workflows exhibit different characteristics e.g., topological structure, size and computation-communication demands while they can have different or conflicting optimization goals. The above results in scheduling decisions of high complexity which in turn may adversely affect the quality of solutions. In this paper we present a fast scheduling algorithm (i.e., the Multiple Workflow Complementary Packing algorithm, MWCP algorithm) for the management of multiple workflows. MWCP combines list scheduling methodologies and Bin Packing techniques to minimize the overall execution time of the workflows. For each workflow the scheduler decides on the best policy considering only the information provided by the workflow in question. We evaluate the performance of the proposed algorithm using real workflow applications being tested under different system heterogeneity levels. Results indicate that performance gains over existing studies are up to 9% while different workflow characteristics reveal different trade-offs on the performance of MWCP.
Oikonomou, P., Kolomvatsos, K., Anagnostopoulos, C., Tziritas, N., Theodoropoulos, G., 'A Probabilistic Batch Oriented Proactive Workflow Management', in , 33rd International Conference on Tools with Artificial Intelligence (ICTAI 2021),November 1-3 (Virtually), 2021.
Abstract

×
Workflow management is a widely studied research subject due to its criticality for the ecient execution of various processing activities towards concluding innovative applications. A set of models has been already proposed dealing with nding the most appropriate node to conclude the placement of each task present in a workflow. The ultimate goal is to eliminate the required time for delivering the nal outcome taking into consideration the dependencies between tasks. In this paper, we go a step forward and enhance the decision making of a scheduler with a batch oriented approach to deal with a high number of workflows. We also focus on a gap of the respective literature, i.e., apart from the time and cost requirements, we focus on the statistics of the underlying data where tasks should be executed. We provide a probabilistic data oriented approach combined with a infrastructure oriented scheme to pay attention on dynamic environments where the underlying data are continuously updated trying to minimize the network overhead for migrating data. We propose the sequential management of workflows, i.e., we map the workflows requirements for data with the available datasets, then, combine the outcome with an optimization model upon the time requirements and the cost of every placement. The performance of our sequential management is revealed by a high number of experiments depicting the advantages in the network overhead. Our evaluation deals with a high number of real workflow applications and a comparative assessment with other baseline schemes.
Fountas, P., Kolomvatsos, K., 'Proactive, Correlation Based Anomaly Detection at the Edge', in , 33rd International Conference on Tools with Artificial Intelligence (ICTAI 2021),November 1-3 (Virtually), 2021.
Abstract

×
Data management at the edge of the network is a significant research subject. Devices present at the Internet of Things (IoT) infrastructure can collect data and transfer them to a set of edge nodes for further processing. There, various activities can be realized. Among them, of great importance it is the detection of anomalies in the incoming data and their preparation to be the subject of the core processing tasks. In this paper, we propose an ensemble scheme for data anomalies detection and elaborate on the use of an extended sliding window approach. We differentiate from the state of the art solutions and argue on the concept of potential anomalies and confirm their presence by incorporating more data into our decision mechanism when it is necessary. The performance of the proposed scheme is evaluated by a set of experimental scenarios and exposed by the provision of numerical results.
Polycarpou, O., Anagnostopoulos, C., Kolomvatsos, K., 'Optimal Load-Aware Task Offloading in Mobile Edge Computing', in 8th International Conference of Control, Dynamic Systems, and Robotics (CDSR 21), May 23-25, Niagara Falls, Canada (Virtually), 2021.
Abstract

×
Mobile Edge Computing (MEC) has emerged as a new computing paradigm to provide computing resources and storing applications closer to the end-users at the operator network boundary. One of the main challenges of MEC is task offloading, i.e., the transfer of computational tasks to a remote processor or external platforms such as a grid of servers or the Cloud. Task offloading mainly faces when and where is best to offload tasks to mitigate a smart device's energy consumption and workload. This paper tackles this challenge by adopting the principles of Optimal Stopping Theory (OST) with three time-optimised sequential decision-making models. A performance evaluation is provided with upon real data-sets on which our proposed models are applied and compared to the theoretical optimal model. Our results show how close our models can be to the theoretical optimal one based on probabilistic and scaling factors. Moreover, in our performance evaluation section, we conclude that one of the applied sequential models can be extremely close to the optimal one making it suitable in single-user and competitive user scenarios.
Koukaras, T., Kolomvatsos, K., 'Proactive Data Allocation in Distributed Datasets based on an Ensemble Model', in 12th International Conference on Information and Communication Systems (ICICS 2021), 24-26 May, Valencia - Spain (Virtual), 2021.
Abstract

×
The advent of the Internet of Things (IoT) gives the opportunity to numerous devices to interact with their environment, collect and process data. Data are transferred, in an upwards mode, to the Cloud through the Edge Computing (EC) infrastructure. A high number of EC nodes become the hosts of distributed datasets where various processing activities can be realized. In this paper, we focus on a model that proactively decides where the collected data should be stored in order to maximize the accuracy of datasets present at the EC infrastructure. We consider that the accuracy is defined by the solidity of datasets exposed as the statistical resemblance of data. We argue upon the similarity of the incoming data with the available datasets and select the most appropriate of them to store the new information. For alleviating processing nodes from the burden of a continuous, complicated statistical processing, we propose the use of synopses as the subject of the similarity process. The incoming data are matched against the available synopses based on an ensemble scheme, then, we select the appropriate host to store them and perform the update of the corresponding synopsis. We provide the description of the problem and the formulation of our solution. Our experimental evaluation targets to reveal the performance of the proposed approach.
Zioviris, G., Kolomvatsos, K., Stamoulis, G., 'On the Use of a Sequential Deep Learning Scheme for Financial Fraud Detection', Computing Conference, , July 15-16, London, 2021.
Abstract

×
Forecasting fraud detection has never been more essential for the finance industry than today. The detection of fraud has been a major concern for the banking industry due to the high impact on banks' revenues and reputation. Fraud can be related with an augmented financial risk, which is often underestimated until it is too late. Recently, deep learning models have been introduced to detect and forecast possible fraud transactions with increased efficiency compared to the conventional machine learning methods and statistics. Such methods gain significant popularity due to their ability to estimate the unknown distribution of the collected data, thus, increasing their capability of detecting more complex fraud events. In this paper, we introduce a novel multistage deep learning model that combines a feature selection process upon an Autoencoder model and a deep convolutional neural network to detect frauds. We plan to adopt the proposed model for performing the `first' stage fraud detection analysis before data are transferred to the Cloud back end for further processing. To manage highly unbalanced datasets, we rely on the Synthetic Minority Over-sampling Technique (SMOTE) to oversample our dataset and adjust the class distribution delivering an efficient classification approach. We describe the problem under consideration and our contribution that provides a solution for it. An extensive set of experimental scenarios are adopted to reveal the performance of the proposed scheme exposing the relevant numerical results. A comparative assessment is used for proving the superiority of our model compared with a Support Vector Machine (SVM) scheme, a classical CNN model and the results of two researches that use the same dataset.
Oikonomou, P., Kolomvatsos, K., Loukopoulos, T., 'Resource Provisioning Schemes for Multiple Workflow Scheduling with Seclusion Requirements', in 24th Pan-Hellenic Conference on Informatics, Nov. 20-22, Athens, Greece, 2020.
Abstract

×
T. Tziouvaras, K. Kolomvatsos, 'Intelligent Monitoring of Virtualized Services', in 8th European Conference on Service-Oriented and Cloud Computing, EU Projects track, 2020.
Abstract

×
Interactive TV applications impose novel requirements in future networks due to the huge volumes of data transferred through the network. ENFORCE provides a framework for the management of such applications targeting to support the real time adaptation on end users needs. The project offers a set of functionalities over virtualized resources as provided by the SoftFIRE platform. Apart from the envisioned functionalities, ENFORCE acts as a benchmarking tool and an extension to the SoftFIRE framework enhancing its monitoring capabilities for the provided virtualized resources and services. Our aim is to pro-actively respond to changes in iTV demand, thus, making the platform fully aligned with the real needs of end users. In ENFORCE, virtual resources are defined for realizing Set top Boxes (STBs) functionalities to be transformed to virtual STBs (vSTBs).
Tsoukas, V., Kolomvatsos, K., Chioktour, V., Kakarountas, A., 'A Comparative Assessment of Machine Learning Algorithms for Events Detection', in 4th South-East Europe Design Automation, Computer Engineering, Computer Networks and Social Media Conference (SEEDA-CECNSM), 2019.
Abstract

×
Nowadays, one can observe massive amount of data production by numerous devices interacting with their environment and end users. Such data can be the subject of advanced processing usually through machine learning algorithms. Hence, we are able to provide intelligent applications and analytics in many research domains like health informatics, information technology, environmental sciences, and so on so forth. However, choosing the appropriate machine learning model for data processing can be one of the most difficult tasks. In this paper, we try to facilitate researchers providing a 'benchmarking' of multiple machine learning algorithms to reveal their advantages and drawbacks. This effort mostly focuses on the accuracy of the studied algorithms and adopts various datasets found in the respective literature. We provide a short description of the adopted models, the datasets and extensive experimental evaluation accompanied by numerical results and our qualitative review on the outcomes.
Panagou, N., Koziri, M., Papadopoulos, P., Oikonomou, P., Tziritas, N., Kolomvatsos, K., Loukopoulos, T., Khan, S., 'Evaluation of Heterogeneous Scheduling Algorithms for Wavefront and Tile Parallelism in Video Coding', in International Conference in Internet of Things (ICIOT) , 2019.
Abstract

×
Video is by far the biggest Big Data, stretching network and storage capacity to their limits. To handle the situation, video compression has been an active field of study for many years, producing output of huge commercial interest, e.g., MPEG-2 and DVD. However, video coding is a computationally expensive process and for this reason, parallelization was proposed at various granularity levels. Of particular interest, are block level methods implemented in HEVC (High Efficiency Video Coding) which was designed to be the successor of H.264/AVC for the 4K era. Parallelization in HEVC is supported by the following three modes: slices, tiles and wavefront. While considerable research was conducted on the parallelization options of HEVC, it was focused on the case of homogeneous processors. In this paper we consider video coding parallelization when the processing elements are heterogeneous. In particular, we focus on wavefront and tile parallelism and measure the performance of scheduling schemes for the induced subtasks. Through simulation experiments with dataset values obtained from common benchmark sequences, we conclude on the relevant merits of the evaluated scheduling algorithms.
P. Papadopoulos, N. Panagou, M. Koziri, K. Kolomvatsos, T. Loukopoulos, I. Anagnostopoulos, 'Coding Time Prediction in H.264/HVEC Transcoding Using Macroblock Sizes', in 14th International Workshop on Semantic and Social Media Adaptation and Personalization, in conjunction with the 27th ACM Conference on User Modeling, Adaptation and Personalization,3-12 June, Larnaca, Cyprus, 2019.
Abstract

×
The continuous customers' demand for higher resolution video led to the development of video coding standards that surpass the limitations of H.264/AVC. Prominent examples in this category include the High Efficiency Video Coding (HEVC) and AV1 which both nominally target at least 50% better compression rate compared to H.264/AVC. However, with a plethora of videos already existing in H.264/AVC format and the majority of cameras in use supporting the older standard, the need for efficient transcoding of H.264/AVC videos into newer standards, cannot be overlooked. Although a significant amount of research exists on fast transcoding of H.264/AVC sequences, most of the works have focused on speeding up block level encoding, usually by taking advantage of mode and motion information that is already contained in the compressed input stream. Nevertheless, equally important to reducing per block transcoding time, is the exploitation of parallelism. Towards this end, identifying in advance the most computationally demanding regions is a prerequisite for efficient resource allocation, especially in a heterogeneous computing environment. In this paper we study the potential of using Macroblock size to predict the most computationally demanding frame regions. Experiments with commonly used benchmark video sequences illustrate the merits of our approach.
P. Oikonomou, A. Dadaliaris, K. Kolomvatsos, T. Loukopoulos, A. Kakarountas, G. Stamoulis, 'Improved Parallel Legalization Schemes for Standard Cell Placement with Obstacles', Technologies, MPDI, 2019.
Abstract

×
In standard cell placement, a circuit is given consisting of cells with a standard height, (different widths) and the problem is to place the cells in the standard rows of a chip area so that no overlaps occur and some target function is optimized. The process is usually split into at least two phases. In a first pass, a global placement algorithm distributes the cells across the circuit area, while in the second step, a legalization algorithm aligns the cells to the standard rows of the power grid and alleviates any overlaps. While a few legalization schemes have been proposed in the past for the basic problem formulation, few obstacle-aware extensions exist. Furthermore, they usually provide extreme trade-offs between time performance and optimization efficiency. In this paper, we focus on the legalization step, in the presence of pre-allocated modules acting as obstacles. We extend two known algorithmic approaches, namely Tetris and Abacus, so that they become obstacle-aware. Furthermore, we propose a parallelization scheme to tackle the computational complexity. The experiments illustrate that the proposed parallelization method achieves a good scalability, while it also efficiently prunes the search space resulting in a superlinear speedup. Furthermore, this time performance comes at only a small cost (sometimes even improvement) concerning the typical optimization metrics.
K. Kolomvatsos, K. Panagidi, I. Neokosmidis, D. Varoutas, S Hadjiefthymiades, ’Automated Concurrent Negotiations: An Artificial Bee Colony Approach’, Elsevier Electronic Research and Applications (ECRA), vol. 19, 2016, pp. 56-69.
Abstract

×
In Electronic Marketplaces (EMs), a number of unknown entities can interact in order to conclude purchase actions. Interactions are between buyers and sellers. Both groups of entities (e.g., buyers, sellers) aim to acquire items in the most profitable price. The discussed interactions are, usually, realized in the form of negotiations over a number of item characteristics. In this paper, we focus on the buyer side and deal with automated multi-issue concurrent negotiations. Such negotiations are between buyers and multiple sellers having in their property specific items. Each buyer negotiates with a number of sellers trying to achieve the most profitable values for a number of items characteristics. We propose an optimization model for achieving the maximum possible utility. Our method adopts the principles of the widely known Artificial Bee Colony (ABC) algorithm that offers a number of advantages compared to other Swarm Intelligence (SI) algorithms (e.g., Particle Swarm Optimization). The buyer, through a number of threads, tries to find the optimal solution concerning the best agreement when negotiating with a group of sellers. Each thread utilizes a weights adaptation scheme for optimizing the utility value. A number of experiments reveal that the proposed model performs better than other efforts found in the literature.
Jankowski, J., Goode, S., Kolomvatsos, K., Kazienko, P., Watrobski, J., ‘Modeling User Preferences and Behaviors in Virtual Product Retail Systems’, Special Issue on Analyzing and Mining Social Networks for Decision Support, Journal of Universal Computer Science, vol. 22(3), pp. 416-437, 2016.
Abstract

×
An important aspect of managing social platforms, online games and virtual worlds is the analysis of user characteristics related to subscriptions and virtual goods purchase. The results of such a process could be adopted in decision support applications that build on top of users' behavior to provide efficient strategies for the virtual world's management. One of the research questions in this area is related to the factors affecting purchases and their relation to the activity within social networks as well as the ability to use past data to make reasoning about future behaviors. Complex online systems are hard to analyze when adopting legacy methodologies due to the huge amount of data generated by users' activity and changes in their behavior over time. In this paper, we discuss an analysis of the characteristics of users performing purchases for virtual products. We adopt a Neuro-Fuzzy system which has the ability to process data under uncertainty towards better decisions related to parameterization of the virtual retail system. The proposed Fuzzy Logic (FL) inference model focuses on the analysis of purchases based on the types of past transactions and social activity as inputs. The proposed system results values for specific parameters affecting/depicting users' behavior like own purchases, gifting and virtual products usage as output. Our results could be adopted for decision support of online platform operators and show the relations between less and more experienced users in terms of frequency and value of purchases, engagement with the use of virtual goods and gifting behaviors. Models based on the social activity with distinguished inbound and outbound social connections show increased interest in virtual goods among users with a higher number of inbound connections as a possible tool for building social position.
K. Kolomvatsos, K. Panagidi, S. Hadjiefthymiades, 'A Load Balancing Module for Post Emergency Management', Elsevier Expert Systems with Applications (ESWA), (IF 2013: 1.965), vol. 42(1), 2014, pp. 657 - 667, 2015.
Abstract

×
The research society has developed a number of models and tools to support emergency management. The proposed models are mainly designed for indoor applications oriented to provide guidance directly to people in danger. Only a few of them deal with outdoor scenarios as well as with providing directions to field commanders or rescue teams. Additionally, load balancing techniques for the optimal allocation of a number of entities into a number of resources are understudied creating a gap in the corresponding research. In this paper, we propose a load balancing model oriented to assist field commanders and rescue teams in a post-emergency scenario. The proposed system could be applied either for indoor or outdoor applications. The module builds on top of the solution provided for the known Santa Fe Bar Problem (SFBP). It consists of an intelligent technique aiming to distribute a number of entities into a finite number of resources. A set of predictors undertake the responsibility of estimating the load of each resource. These predictors are adopted to select the appropriate resource for each entity. A case study deals with the distribution of injured persons into a number of hospitals and presents the functionality of the proposed module. Finally, numerical results reveal computational and time requirements of our system.
K. Panagidi, K. Kolomvatsos, S. Hadjiefthymiades, ‘An Intelligent Scheme for Concurrent Multi-Issue Negotiations’, International Journal of Artificial Intelligence (IJAI), vol. 12(1), 2014, pp. 129-149.
Abstract

×
Automated negotiations are an active research field for many years. In negotiations, participants' characteristics play a crucial role to the final result. The most important characteristics are the deadline and the strategy. The deadline defines the time for which each entity will participate in the negotiation while the strategy defines the proposed prices at every round. We focus on the buyer side. We study multi-issue concurrent negotiations between a buyer and a set of sellers. In this setting, the buyer utilizes a number of threads. We propose the use of known optimization techniques for updating the buyer behavior as well as a method based on the known Particle Swarm Optimization (PSO) algorithm for threads coordination. The PSO algorithm is used to lead the buyer to the optimal solution (best deal) through threads team work. This way, we are able to provide an efficient mechanism for decision making in the buyer's side. In real situations, there is absolutely no knowledge on the characteristics of the involved entities. In this paper, we combine the proposed methods with the Kernel Density Estimator (KDE) and Fuzzy Logic (FL) in order to handle incomplete knowledge on entities characteristics. When an agreement is true in the set of threads, KDE is responsible to give to the rest of them the opportunity to calculate the probability of having a better agreement or not. This result is fed to a FL controller in order to adapt the behavior of each thread. Our experimental results depict the efficiency of the proposed techniques through results for known evaluation parameters.
K. Kolomvatsos, S. Hadjiefthymiades, 'On the Use of Particle Swarm Optimization and Kernel Density Estimator in Concurrent Negotiations', accepted for publication in Elsevier Information Sciences (INS) (IF 2013: 3.893), 2014.
Abstract

×
Electronic Marketplaces (EMs) can offer a number of advantages for users searching for products. In EMs, Intelligent Agents (IAs) can undertake the responsibility of representing buyers and sellers and negotiate over the conclusion of purchases. For this purpose, a negotiation is held between IAs. In real situations, there is absolutely no knowledge for the entities characteristics. The most important characteristics are the deadline and the pricing strategy. The strategy defines the proposed prices at every round of the negotiation. In this paper, we model such kind of uncertainty by utilizing known techniques for estimating the deadlines and strategies distributions. One of them is the Kernel Density Estimation (KDE) technique. We focus on the buyer side. We study concurrent negotiations between a buyer and a set of sellers. For this, the buyer utilizes a number of threads. Each thread follows a specific strategy and adopts swarm intelligence techniques for achieving the best agreement. Particle Swarm Optimization algorithm is adopted by each thread. The most important is that in our setting there is no need for a coordinator. The buyer can select upon the best agreement among the set of sellers. Our experimental results depict the time interval where the agreement is possible and the efficiency of the proposed model.
K. Kolomvatsos, C. Anagnostopoulos, and S. Hadjiefthymiades, ‘Determining the Optimal Stopping Time for Automated Negotiations’, IEEE Transactions on Systems, Man and Cybernetics (Part: Systems), doi 10.1109/TSMC.2013.2279665, 2013.
Abstract

×
Electronic Markets (EMs) are virtual frameworks where entities not known in advance have the opportunity to interact for the trading of products or services. Usually, a negotiation is necessary for the conclusion of the transaction. The conclusion is either positive (agreement) or negative (conflict). An efficient reasoning mechanism is necessary for players participating in negotiations. In this paper, we focus on the buyer side and propose two decision models based on the Optimal Stopping Theory (OST). OST is proved to be very efficient in cases where an entity tries to find the time to stop a process with the aim of maximizing her utility. The outcome of the proposed decision method indicates whether the buyer stops a negotiation either by accepting the offer or continuing in the negotiation by rejecting it. In our models, we assume zero knowledge on the players' characteristics. Our proposed decision models do not require any complex modelling or any information provided by experts. Experimental results reveal the efficiency of each model and provide a comparison assessment with other research efforts.
Kostas Kolomvatsos and Stathes Hadjiefthymiades, 'An Extended Q-Gram Algorithm for Calculating the Relevance Factor of Products in Electronic Markets', Elsevier Electronic Commerce Research and Applications (ECRA) (IF 2013: 1.304), vol. 12(6), 2013, pp. 397-411.
Abstract

×
Intelligent Agents (IAs) can offer a number of advantages when used in Electronic Markets (EMs). In such environments, IAs can represent users that act as buyers or sellers. In the buyer's side, an IA could undertake the responsibility of finding and purchasing products that meet the owner's needs. In this process, the IA should decide if a product, offered by a seller, is relevant to the owner's preferences. We propose an algorithm for calculating the relevance factor of a product based on the product description, constraints defined by the buyer and the product's Quality of Service (QoS) characteristics (such as the delivery time, the seller trust level). The proposed algorithm is based on widely known similarity assessment techniques. However, we also propose a new similarity assessment scheme based on the Q-grams technique. We describe the proposed solution and evaluate our methodology. Results show that the algorithm is an efficient way for the relevance factor calculation and QoS characteristics play an important role in the calculation process. QoS factor calculation provides an additional level of intelligence in the proposed methodology.
T. Loukopoulos, M. Koziri, K. Kolomvatsos, P. Oikonomou, 'On Green Scheduling for Desktop Grids', to be presented in Pervasive Information Systems Workshop, 6th International Conference on Information Systems and Technologies, Naples, Italy, March 27-29, 2018.
Abstract

×
Task scheduling is of paramount importance in a desktop grid environment. Earlier works in the area focused on issues such as: meeting task deadlines, minimizing make-span, monitoring and checkpointing for progress, malicious or erroneous peer discovery and fault tolerance using task replication. More recently energy consumption has been studied from the standpoint of judiciously replicating and assigning tasks to the more power efficient peers. In this paper we tackle another aspect of power efficiency with regards to scheduling, namely greenness of the consumed energy. We give a formulation as a multi-objective optimization problem and propose heuristics to solve it. All the heuristics are evaluated via simulation experiments and conclusions on their merits are drawn.
Panagiota Papadopoulou, Kostas Kolomvatsos, Kyriaki Panagidi, Stathes Hadjiefthymiades, 'Investigating the Business Potential of Internet of Things', in Proceedings of the Mediterranean Conference on Information Systems, Sept. 4-6, Genoa, Italy, 2017.
Abstract

×
Internet of Things (IoT) encompasses a wide range of devices and technologies which cumulatively shape a new environment with unprecedented business prospects. This paper aims to investigate the business potential of the IoT, examining the opportunities it offers as well as the challenges it creates for current and future business models. In this direction, it presents a proposed framework for analyzing IoT business models into dimensions that can facilitate our understanding of their development and success. The paper also denotes factors that can affect the success of IoT business models, focusing on security, privacy, trust, legal and economic aspects of IoT. The paper continues to show the business applicability of IoT through the case of RAWFIE project, describing example application scenarios of mobile IoT in selected domains and analysing its business potential.
K. Panagidi, K. Kolomvatsos and S. Hadjiefthymiades, 'On the Use of PSO with Weights Adaptation in Concurrent Multi-Issue Negotiations', in 10th International Symposium on Distributed Computing and Artificial Intelligence (DCAI '13), Salamanca, Spain, May 22-24, 2013.
Abstract

×
In this paper, we deal with automated multi-issue concurrent negotiations. A buyer utilizes a number of threads for negotiating with a number of sellers. We propose a method based on the known PSO algorithm for threads coordination. The PSO algorithm is used to lead the buyer to the optimal solution (best deal) through threads team work. Moreover, we propose a weights adaptation scheme for optimizing buyer behavior and promoting efficiency. This way, we are able to provide an efficient mechanism for decision making in the buyer's side. This is proved by our results through a wide range of experiments.
K. Kolomvatsos, K. Panagidi and S. Hadjiefthymiades, 'Optimal Spatial Partitioning for Resource Allocation', in 10th International Conference on Information Systems for Crisis Response and Management (ISCRAM 2013), Baden-Baden, Germany, May 12-15, 2013.
Abstract

×
Spatial partitioning consists of the problem of finding the best segmentation of an area under specific conditions. The final goal is to identify parts of the area where a number of resources could be allocated. Such cases are common in disaster management scenarios. In this paper, we consider such a scenario and propose a methodology for the resource allocation for emergency response. We utilize an intelligent technique that is based on the Particle Swarm Optimization algorithm. We define the problem by giving specific formulations and describe the proposed algorithm. Moreover, we provide a method for separating the area into cells and describe a technique for calculating cell weights based on the underlying spatial data. Finally, we present a case study for allocating a number of ambulances and give numerical results concerning the run time and the total coverage of the examined area.
G. Boulougaris, K. Kolomvatsos, and S. Hadjiefthymiades, 'Building the Knowledge Base of a Buyer Agent Using Reinforcement Learning Techniques', In Proc. of the 2010 IEEE World Congress on Computational Intelligence (WCCI 2010), IJCNN, July 18th - 23rd, Barcelona, Spain, pp. 1166-1173.
Abstract

×
Electronic markets are places where entities not known in advance can negotiate and agree upon the exchange of products. Intelligent agents can be proved very advantageous when representing entities in markets. Mostly, such entities are based on reputation models in order to conclude a transaction. However, reputation is not the only parameter that they could be based on. In this work, we deal with the problem of how and on which entity a buyer should be rely upon in order to conclude a transaction. Reinforcement learning techniques are used for these purposes. More specifically, the Q-learning algorithm is used for the calculation of the reward that the buyer will take for every action in the market environment. Actions represent the selection of specific entities for the negotiation of products. The most important is that the reward values are calculated based on a number of parameters such as the price, the delivery time, etc. The result is a more efficient model that is not based only on the reputation of each entity. Finally, we extend the Q-learning algorithm and propose a methodology for the dynamic Q-table creation which results reduced time for its construction and respectively limited time for the purchase action. Simulations show that this model indicates a significant time reduction in the purchase process in conjunction with the best solution according to the characteristics of products.
K. Kolomvatsos and S. Hadjiefthymiades, 'Implicit Deadline Calculation for Seller Agent Bagraining in Information Marketplaces', In Proc. of the 2nd International Conference on Complex, Intelligent and Software Intensive Systems (CISIS 2008), March 4th - 7th, Polytechnic University of Catalonia, Barcelona, Spain, 2008, pp. 184-190.
Abstract

×
Present and future Web business models involve the trading of information goods. Information marketplaces can be considered as places where users search and retrieve information goods. Such places appear to be very interesting information retrieval models. Furthermore, Software Agent technology could help users and providers to work in such open environments providing a variety of advantages. Users as well as information providers could be represented by intelligent agents that work autonomously. The representatives of users assume the role of information buyers while the representatives of information sources could be referred to as sellers. In this paper, we examine a scenario where agents representing entities involved in an information marketplace bargain over the prices of information goods. Bargaining originates in Game Theory (GT). The rationale is that some entities contest to gain as much profit as possible in an open environment. We study the sellers' side. Sellers involved in a number of games with buyers, are trying to achieve as greater prices as possible in order to gain more profit from each game. We present a theoretical model of deadline computation for which sellers are participating in the game. Over this time limit it is useless for sellers to continue the game while buyers reject the proposed prices.
Kostas Kolomvatsos and Stathes Hadjiefthymiades, 'Defining Time Constraints for Sellers in Electronic Markets', in the 'Encyclopedia of E-Business Development and Management in the Global Economy', ed. In Lee, IGI Global, 2010.
Abstract

×
Electronic marketplaces provide places for negotiation over the exchange of products. In such places entities representing buyers and sellers can interact and agree upon a product price. Both parties try to maximize their profit. We model such interaction as a finite horizon bargaining game and try to quantify the maximum time of seller participation in the game. The estimated deadline indicates until when the interaction in the game is profitable for the seller. Our model defines the appropriate value for a patience factor which finally results the seller deadline fully adapted in each product characteristics.
Kostas Kolomvatsos and Stathes Hadjiefthymiades, 'How Can We Trust Agents in Multi-Agent Environments?', Chapter in 'Intelligence Integration in Distributed Knowledge Management', eds D. Krol and N. T. Nguyen, IDEA Group Inc., 2008.
Abstract

×
The field of Multi-Agent Systems (MAS) is an active area for many years due to the importance that agents have to many disciplines of research in Computer Science. MAS are open and dynamic systems where a number of autonomous software components, called agents, communicate and cooperate in order to achieve their goals. In such systems, trust plays an important role. There must be a way for an agent to make sure that it can trust another entity, which is a potential partner. Without trust, agents cannot cooperate effectively and without cooperation they cannot fulfill their goals. Many times, trust is based on reputation. It is an indication that we may trust someone. This important research area is investigated in this book chapter. We discuss main issues concerning reputation and trust in MAS. We present research efforts and give formalizations useful for understanding the two concepts.

Semantic Technologies and Domain Specific Languages

Kostas Kolomvatsos, George Valkanas, and Stathes Hadjiefthymiades, 'Debugging Applications Created by a Domain Specific Language: The IPAC Case', Elsevier Journal of Systems and Software (JSS) (IF 2012: 1.135), vol. 85(4), April 2012, pp. 932-943.
Abstract

×
Nowadays, software developers have created a large number of applications in various research domains of Computer Science. However, not all of them are familiar with the majority of the research domains. Hence, Domain Specific Languages (DSLs) can provide an abstract, concrete description of a domain in terms that can easily be managed by developers. The most important in such cases is the provision of a debugger for debugging the generated software based on a specific DSL. In this paper, we propose and present a simple but efficient debugger created for the needs of the IPAC system. The debugger is able to provide debugging facilities to developers that define applications for autonomous mobile nodes. The debugger can map code lines between the initial application workflow and the final code defined in a known programming language. Finally, we propose a logging server responsible to provide debugging facilities for the IPAC framework. The IPAC system is consisted by a number of middleware services for mobile nodes acting in a network. In this system a number of mobile nodes exchanged messages that are visualized for more efficient manipulation.
V. Nomikos, K. Kolomvatsos, and V. Papadopoulos, 'An Application Creation Environment for Autonomic Computing', In 2nd Student Workshop on Wireless Sensor Networks, Oct. 30-31st, Athens, 2009.
Abstract

×
The IPAC Application Creation Environment (ACE) is an Integrated Development Environment (IDE) solution, essentially a tool, aiming to provide all the necessary facilities for creating and editing applications deployable on the mobile IPAC nodes. It is based on the Eclipse platform and follows a model-driven architecture approach. A set of easy to use tools are available to the IPAC developer for editing and testing purposes. All such tools are built around a Domain Specific Language (DSL) designed and implemented for the purposes of the project. Applications can be created using either a textual or a visual editor and can be tested using debugging and emulation tools.
K. Kolomvatsos, V. Papataxiarhis, and S. Hadjiefthymiades, 'Semantic Location Based Services for Smart Spaces', In Proc. of the 2nd International Conference on Metadata and Semantics Research (MTSR), Oct. 11th - 12th, Corfu, Greece, 2007.
Abstract

×
Enhancing the physical environment of users with IT and communication elements is one of the main objectives of the pervasive computing paradigm. The so-called smart spaces , which are typical pervasive computing environments, combine computing infrastructure with intelligent and context-aware services in order to advance the users' computing experience. In this paper we describe a metadata-based infrastructure that is required for delivering semantics-aware location-based services in smart spaces. This infrastructure involves geometric and ontological spatial representation as well as graph- and knowledge-based navigation algorithms.
K. Kolomvatsos, M. Tsiroukis, S. Hadjiefthymiades, ‘An Experiment Description Language for Supporting Mobile IoT Applications’, 2016 FIRE Book, European Commission, River Publishers.
Abstract

×
Mobile IoT applications consist of an innovative field where numerous devices collect, process and exchange huge amounts of data with central systems or their peers. Intelligent applications could be built on top of such information realizing the basis of future Internet. For engineering novel applications, experimentation plays a significant role, especially, when it is performed remotely. Remote experimentation should offer a framework where experimenters can efficiently define their experiments. In this paper, we focus on the experiments definition management proposed by Road-, Air- and Water-based Future Internet Experimentation (RAWFIE). RAWFIE offers, among others, an experimentation language and an editor where experimenters can remotely insert their experiments to define actions performed by the nodes in a testbed. RAWFIE proposes the Experiment Description Language (EDL) that provides the elements for the management of devices and the collected data. Commands related to any aspect of a node behavior (e.g., configuration, location, task description) are available to experimenters. We report on the EDL description and the offered editors and discuss their key parts and functionalities.
Kostas Kolomvatsos, George Valkanas, Petros Patelis, and Stathes Hadjiefthymiades, 'Creating, Debugging and Testing Mobile Applications with the IPAC Application Creation Environment', in 'Formal and Practical Aspects of Domain-Specific Languages: Recent Developments', ed. Dr M. Mernik, IGI Global, 2012.
Abstract

×
An important challenge in software development is to have efficient tools for creating, debugging and testing software components developed for specific business domains. This is more imperative if we consider that a large number of users are not familiar with popular programming languages. Hence, Application Creation Environments (ACEs) based on specific Domain-Specific Languages (DSLs) can provide an efficient way for creating applications for a specific domain of interest. The provided ACEs should incorporate all the functionality needed by developers to build, debug and test applications. In this chapter, we present our contribution in this domain based on the experience of the IPAC system. The IPAC system provides a middleware and an ACE for developing and using intelligent, context-aware services in mobile nodes. We fully describe the ACE which is a key part of the overall architecture. Our ACE provides two editors (textual, visual), a wide functionality spectrum as well as a debugger and an application emulator. The ACE is based on an Application Description Language (ADL) developed for IPAC. The ADL provides elements for the description of an application workflow for embedded systems. Through such functionality, developers are capable of efficiently creating and testing applications that will be deployed on mobile nodes.
Kostas Kolomvatsos and Stathes Hadjiefthymiades, 'Ontologies and Intelligent Agents: A Powerful Bond', Chapter in 'The Semantic Web for Knowledge and Data Management: Technologies and Practices', eds Z. Ma and H. Wang, IDEA Group Inc., 2008.
Abstract

×
The emerged form of information with computer-processable meaning (semantics) as presented in the framework of the Semantic Web (SW) facilitates machines to access it more efficiently. Information is semantically annotated in order to ease the discovery and retrieval of knowledge. Ontologies are the basic element of the SW. They carry knowledge about a domain and enable interoperability between different resources. Another technology that draws considerable attention nowadays (shows major interest, especially today), is the technology of Intelligent Agents. Intelligent agents act on behalf of a user to complete tasks and may adapt their behavior to achieve their objectives. The objective of this Chapter is to provide an exhaustive description of fundamentals regarding the combination of SW and intelligent agent technologies.
Kostas Kolomvatsos, 'Ubiquitous Computing in Education', Chapter in 'Ubiquitous and Pervasive Knowledge anD Learning Management: Semantics, Social Networking and New Media to their Full Potential', IDEA Group Inc., 2007.
Abstract

×
With the development of technology, new roads have been opened in Education. An interesting idea is to use the computers in teaching and learning procedure. Students will have the opportunity to gain access to information resources in a timeless and limitless way. Teachers will be able to transform their classes in a student-centered environment avoiding the drawbacks that the traditional teacher-centered model has. In this direction ubiquitous computing has significant advantages. Ubiquitous means that computational devices are distributed into physical world giving us boundless access to communication and information channels. Now, knowledge can be built based on collaboration, communication, experimentation and on students' experiences. Research has shown positive impacts on learning. This chapter deals with issues directly connected to ubiquitous computing, such as its features, types of devices used, and pedagogical goals. The advantages and disadvantages of ubiquitous environments are fully examined and some initiatives are referred.
V. Nomikos, and K. Kolomvatsos, 'Documentation of the IPAC Application Description Language', Technical Report, Department of Informatics and Telecommunications, National and Kapodistrian Univarsity of Athens, Sept. 2009.
Abstract

×
This document is the updated version of the first release of the Documentation of the IPAC Application Description Language and it is an attempt to describe the syntax and some implementation details of the Application Description Language (ADL). The ADL consists of a set of 42 EBNF?like rules describing structures and elements that the developer will be able to use in order to define a new IPAC Application. Using the openArchitectureWare framework, and more specifically the Xtext framework, we take an auto generated parser and an Ecore meta?model for the definition of the ADL. These components are necessary for the correct definition of an IPAC application and the editors syntactic checking mechanism. Moreover, some extensions are used to enrich our language and the derived metamodel with useful functionalities. For example, we have defined a set of Java classes that provide us methods used for checking purposes or in the content assist component. Just to note one, the method getServices() is used in order to have access in the service model and accordingly to retrieve all the available service names and methods. In the following sections, we try to present the syntax of all the available elements of the ADL and give some practical examples.
K. Kolomvatsos and V. Tsetsos, 'Spatial Ontology-Based Annotation Using GIS', Technical Report, Department of Informatics and Telecommunications, National and Kapodistrian Univarsity of Athens, Feb. 2008.
Abstract

×
This report presents a methodology for spatial annotation based on an ontology and using a Geographical Information System (GIS). The steps required at every stage of the procedure are described. The annotation process involves the use of GIS software with which spatial elements are defined. GISs help developers to map data concerning the location of various spatial elements such as entrances, navigational points (end points, turn points, junctions), transitions points, etc. These points are stored to a spatial database in order to be used features which relate to the processing of data such as the geometry of points, lines, polygons, etc. Finally, the ontology population is held based on the data retrieved from the spatial database.

Other Domains

Axelou, O., Kolomvatsos, K., Floros, G., Evmorfopoulos, N., Georgakos, G., Stamoulis, G., 'An Electromigration-Aware Wire Sizing Methodology via Particle Swarm Optimization', in 34th Great Lakes Symposium on VLSI (GLSVLSI),, June 12-14, Clearwater , FL , USA, 2024.
Abstract

×
Tsolia, M., Zygouris, N., Kolomvatsos, K., 'Neural Networks for Assessing Reading Disabilities in School-Aged Children', in IEEE Global Engineering Education Conference (EDUCON),, May 8-11, Kos, Greece, 2024.
Abstract

×
Reading, or the ability to infer meaning from printed words in order to correctly interpret relevant information, is the most fundamental component of education. To recognize letters, letter strings, and words, one must possess the ability to decode abstract graphemes accurately and fluently into their corresponding phonemes. Additionally, processing text requires the capacity to read and comprehend text with both fluency and accuracy. Thus, reading requires a variety of cognitive abilities, including effective processing speed, phonological awareness, syntactic processing, auditory and visual word recognition, and phonological awareness. Thepresent study reports the outcomes of a research that evaluated the identification of students with reading disabilities using an artificial neural network. The neural network consisted of structured tasks aiming at a) reading, b) distinguishing words and pseudowords and c) reading two texts. Participants were 235 children attending grades from third to sixth class. Audio is converted into a spectrogram and students with disabilities are identified using machine learning algorithms and auditory analysis. The outcome of the present study suggests that a neural network that is comprised from three tasks can identify the reading abilities and classify the school aged children between typical achievers and reading disabled. Furthermore, including the mAP scores, the results show that the model is highly effective in identifying and classifying reading difficulties in real-time, offering a promising avenue for future research and practical applications in educational settings.
Diamantas, A., Terpo, I., Chondros, P., Chatzipetrou, V., Vagenas, S., Papathanasaki, M., Fountas, P., Kolomvatsos, K., 'Digitization and Digital Presentation of a Wooden Traditional Boat', in 5th Pan-Hellenic Conference on Digital Cultural Heritage, Euromed ‘23,, Larissa, Greece, March 6-9, 2024.
Abstract

×
Nowadays, the development of new technologies and methods for digitization facilitates the preservation, highlighting, promotion and presentation of cultural heritage artifacts. It is widely known that many countries at the Mational level, but also organizations such as museums, libraries, etc. own and display repositories of digital cultural content offering remote access to cultural heritage objects and monuments. In addition, the importance of new digitization technologies is easily understood if we consider that many monuments are at risk and should be saved digitally and documented to be offered to future generations. In this article, the processes of digitization and digital documentation of the ELENI P. wooden ship, which was dismantled in a controlled manner as part of the SaveWoodenBoats research project, are presented. The ultimate goal is to present the technologies adopted to digitize every part of the vessel and then the digitized information to be the basis for its presentation through augmented, virtual and mixed reality technologies. We describe the process adopted step-by-step and analyze the technical challenges we encountered during the efforts to digitally highlight ELENI P. The innovative approaches of the project are presented in detail while at the same time we capture examples of the final user application through which one can navigate and to see the construction details of the vessel.
Vavougios, G., Stavrou, V., Konstantatos, C., Sinigalias, P.-C., Zarogiannis, S., Kolomvatsos, K., Stamoulis, G., Gourgoulianis, K., 'COVID-19 Phenotypes and Comorbidity: A Data-Driven, Pattern Recognition Approach Using National Representative Data from the United States', International Journal of Environmental Research and Public Health, 19, 4630, 2022.
Abstract

×
The aim of our study was to determine COVID-19 syndromic phenotypes in a data-driven manner using the survey results based on survey results from Carnegie Mellon University s Delphi Group. Monthly survey results (>1 million responders per month; 320,326 responders with a certain COVID-19 test status and disease duration <30 days were included in this study) were used sequen tially in identifying and validating COVID-19 syndromic phenotypes. Logistic Regression-weighted multiple correspondence analysis (LRW-MCA) was used as a preprocessing procedure, in order to weigh and transform symptoms recorded by the survey to eigenspace coordinates, capturing a total variance of >75%. These scores, along with symptom duration, were subsequently used by the Two Step Clustering algorithm to produce symptom clusters. Post-hoc logistic regression models adjusting for age, gender, and comorbidities and confirmatory linear principal components analyses were used to further explore the data. Model creation, based on August s 66,165 included responders, was subsequently validated in data from March December 2020. Five validated COVID-19 syndromes were identified in August: 1. Afebrile (0%), Non-Coughing (0%), Oligosymptomatic (ANCOS); 2. Febrile (100%) Multisymptomatic (FMS); 3. Afebrile (0%) Coughing (100%) Oligosymptomatic (ACOS); 4. Oligosymptomatic with additional self-described symptoms (100%; OSDS); 5. Olfac tion/Gustatory Impairment Predominant (100%; OGIP). Our findings indicate that the COVID-19 spectrum may be undetectable when applying current disease definitions focusing on respiratory symptoms alone.
K. Kolomvatsos, P. Papadopoulou, S. Hadjiefthymiades, 'Internet of Things Business Models: The RAWFIE Case', in 18th IFIP Conference on e-Business, e-Services, and e-Society, Norway, Sept. 2019.
Abstract

×
Internet of Things (IoT) encompasses a wide range of devices and technologies which cumulatively shape a new environment with unprecedented business prospects. This paper aims to investigate the business potential of IoT, focusing on mobile IoT and IoT as a service. In this direction, it presents the case of RAWFIE, a H2020 project on a research and experimentation federated platform of mobile IoT testbeds and devices. The paper describes RAWFIE potential business models, analyzing them into their characteristics, based on the business model canvas, examining the possibilities as well as the challenges these entail, using a SWOT analysis, and testing them in a preliminary evaluation. The study offers research and practical insights as a starting point for business models of IoT as a service, focusing in the context of mobile IoT experimentation.
P. Papadopoulou, K. Kolomvatsos, S. Hadjiefthymiades, 'Enhancing E-Government with Internet of Things', in 'Computational Intelligence in the Internet of Things', Chapter 5, IGI Global, 2019.
Abstract

×
Internet of things (IoT) brings unprecedented changes to all contexts of our lives, as they can be informed by smart devices and real-time data. Among the various IoT application settings, e-government seems to be one that can be greatly benefited by the use of IoT, transforming and augmenting public services. This chapter aims to contribute to a better understanding of how IoT can be leveraged to enhance e-government. IoT adoption in e-government encompasses several challenges of technical as well as organizational, political, and legal nature, which should be addressed for developing efficient applications. With the application of IoT in e-government being at an early stage, it is imperative to investigate these challenges and the ways they could be tackled. The chapter provides an overview of IoT in e-government across several application domains and explores the aspects that should be considered and managed before it can reach its full potential.

Dr Kostas Kolomvatsos

personal website

Publications per Research Domain

Intelligent Systems and Analytics, Tasks and Data Management in Pervasive & Edge Computing and IoT

Contextual Reasoning

Fuzzy Logic Reasoning and Applications

Recommender Systems

Smart Cities Applications

Autonomous Behaviour and Decision Making

Semantic Technologies and Domain Specific Languages

Other Domains