Basic Filters
2024 |
George Papadopoulos Alevizos Bastas, George Vouros Ian Crook Natalia Andrienko Gennady Andrienko Jose Manuel Cordero A Deep reinforcement learning in service of air traffic controllers to resolve tactical conflicts Journal Article Expert Systems with Applications, 236 , 2024, ISBN: 0957-4174. @article{Papadopoulos2024, title = {Deep reinforcement learning in service of air traffic controllers to resolve tactical conflicts}, author = {George Papadopoulos, Alevizos Bastas, George A. Vouros, Ian Crook, Natalia Andrienko, Gennady Andrienko, Jose Manuel Cordero }, doi = {https://doi.org/10.1016/j.eswa.2023.121234}, isbn = { 0957-4174}, year = {2024}, date = {2024-02-01}, journal = {Expert Systems with Applications}, volume = {236}, abstract = {Dense and complex air traffic requires higher levels of automation than those exhibited by tactical conflict detection and resolution (CD&R) tools that air traffic controllers (ATCOs) use today: AI tools can act on their own initiative, increasing the capacity of ATCOs to control higher volumes of traffic. However, given that the air traffic control (ATC) domain is safety critical, requires AI systems to which ATCOs are comfortable to relinquishing control, guaranteeing operational integrity and automation adoption. Two major factors towards this goal are quality of solutions and operational transparency. ResoLver, the system that this article presents, addresses these challenges using an enhanced graph convolutional reinforcement learning method operating in a multiagent setting where each agent – representing a flight – performs a CD&R task, jointly with other agents. We show that ResoLver can provide high-quality solutions with respect to stakeholders interests (air traffic controllers and airspace users), addressing also operational transparency issues, which have been validated by ATCOs in simulated real-world settings.}, keywords = {}, pubstate = {published}, tppubtype = {article} } Dense and complex air traffic requires higher levels of automation than those exhibited by tactical conflict detection and resolution (CD&R) tools that air traffic controllers (ATCOs) use today: AI tools can act on their own initiative, increasing the capacity of ATCOs to control higher volumes of traffic. However, given that the air traffic control (ATC) domain is safety critical, requires AI systems to which ATCOs are comfortable to relinquishing control, guaranteeing operational integrity and automation adoption. Two major factors towards this goal are quality of solutions and operational transparency. ResoLver, the system that this article presents, addresses these challenges using an enhanced graph convolutional reinforcement learning method operating in a multiagent setting where each agent – representing a flight – performs a CD&R task, jointly with other agents. We show that ResoLver can provide high-quality solutions with respect to stakeholders interests (air traffic controllers and airspace users), addressing also operational transparency issues, which have been validated by ATCOs in simulated real-world settings. |
2023 |
Antonio Gracia-Berna Jose Manuel Cordero, Natividad Valle Ruben Rodriguez Gennady Andrienko Natalia Andrienko George Vouros Ian Crook Sandrine Molton A Framework for transparent, explainable and trustworthy automation of ATM Conference Forthcoming DASC 2023, Forthcoming. @conference{Gracia-Berna´2023, title = {Framework for transparent, explainable and trustworthy automation of ATM}, author = {Antonio Gracia-Berna, Jose Manuel Cordero, Natividad Valle, Ruben Rodriguez, Gennady Andrienko, Natalia Andrienko, George A. Vouros, Ian Crook, Sandrine Molton}, year = {2023}, date = {2023-06-20}, booktitle = {DASC 2023}, abstract = {Scientific studies before the COVID-19 pandemic indicated that Air Traffic Management (ATM) was close to saturation. The integration of Artificial Intelligence (AI) into ATM has been identified as crucial to achieve higher levels of automation, and the need for trustworthy and explainable automation systems in safety-critical domains is essential. Boeing Aerospace Spain (BAS) has conducted pioneering research on achieving high levels of automation while ensuring transparency and explainability. BAS, along with other major ATM players, has developed a framework for implementing transparent and explainable automation using Explainable Artificial Intelligence (XAI) for Air Traffic Flow and Capacity Management (ATFCM), and Conflict Detection and Resolution (CDR) scenarios. The framework provides a set of principles for the transparent application of XAI technology in ATM to ensure that different types of audiences can trust the AI system’s decisions. The principles have been developed based on feedback from experts in ATM, human factors, and AI. This framework can be considered the first attempt to pave the way for AI techniques to achieve higher levels of automation in accordance with the European ATM Master Plan.}, keywords = {}, pubstate = {forthcoming}, tppubtype = {conference} } Scientific studies before the COVID-19 pandemic indicated that Air Traffic Management (ATM) was close to saturation. The integration of Artificial Intelligence (AI) into ATM has been identified as crucial to achieve higher levels of automation, and the need for trustworthy and explainable automation systems in safety-critical domains is essential. Boeing Aerospace Spain (BAS) has conducted pioneering research on achieving high levels of automation while ensuring transparency and explainability. BAS, along with other major ATM players, has developed a framework for implementing transparent and explainable automation using Explainable Artificial Intelligence (XAI) for Air Traffic Flow and Capacity Management (ATFCM), and Conflict Detection and Resolution (CDR) scenarios. The framework provides a set of principles for the transparent application of XAI technology in ATM to ensure that different types of audiences can trust the AI system’s decisions. The principles have been developed based on feedback from experts in ATM, human factors, and AI. This framework can be considered the first attempt to pave the way for AI techniques to achieve higher levels of automation in accordance with the European ATM Master Plan. |
Georgios Santipantakis Christos Doulkeridis, George Vouros A An Ontology for Representing and Querying Semantic Trajectories in the Maritime Domain Conference Forthcoming ADBIS 2023, Forthcoming. @conference{Santipantakis2023, title = {An Ontology for Representing and Querying Semantic Trajectories in the Maritime Domain}, author = {Georgios Santipantakis, Christos Doulkeridis, George A. Vouros}, year = {2023}, date = {2023-06-20}, booktitle = {ADBIS 2023}, keywords = {}, pubstate = {forthcoming}, tppubtype = {conference} } |
Alevizos Bastas, George Vouros A Data-Driven Modeling of Air Traffic Controllers’ Policy to Resolve Conflicts Journal Article Aerospace, 10 (6), 2023, ISSN: 2226-4310. @article{Bastas2023, title = {Data-Driven Modeling of Air Traffic Controllers’ Policy to Resolve Conflicts}, author = {Alevizos Bastas, George A. Vouros}, doi = {https://doi.org/10.3390/aerospace10060557}, issn = {2226-4310}, year = {2023}, date = {2023-06-13}, journal = {Aerospace}, volume = {10}, number = {6}, abstract = {With the aim to enhance automation in conflict detection and resolution (CD&R) tasks in the air traffic management (ATM) domain, this article studies the use of artificial intelligence and machine learning (AI/ML) methods to learn air traffic controllers’ (ATCOs) policy in resolving conflicts among aircraft assessed to violate separation minimum constraints during the en route phase of flights, in the tactical phase of operations. The objective is to model ℎ𝑜𝑤 conflicts are being resolved by ATCOs. Towards this goal, the article formulates the ATCO policy learning problem for conflict resolution, addresses the challenging issue of an inherent lack of information in real-world data, and presents AI/ML methods that learn models of ATCOs’ behavior. The methods are evaluated using real-world datasets. The results show that AI/ML methods can achieve good accuracy on predicting ATCOs’ actions given specific conflicts, revealing the preferences of ATCOs for resolution actions in specific circumstances. However, the high accuracy of predictions is hindered by real-world data-inherent limitations.}, keywords = {}, pubstate = {published}, tppubtype = {article} } With the aim to enhance automation in conflict detection and resolution (CD&R) tasks in the air traffic management (ATM) domain, this article studies the use of artificial intelligence and machine learning (AI/ML) methods to learn air traffic controllers’ (ATCOs) policy in resolving conflicts among aircraft assessed to violate separation minimum constraints during the en route phase of flights, in the tactical phase of operations. The objective is to model ℎ𝑜𝑤 conflicts are being resolved by ATCOs. Towards this goal, the article formulates the ATCO policy learning problem for conflict resolution, addresses the challenging issue of an inherent lack of information in real-world data, and presents AI/ML methods that learn models of ATCOs’ behavior. The methods are evaluated using real-world datasets. The results show that AI/ML methods can achieve good accuracy on predicting ATCOs’ actions given specific conflicts, revealing the preferences of ATCOs for resolution actions in specific circumstances. However, the high accuracy of predictions is hindered by real-world data-inherent limitations. |
Andreas Kontogiannis, George Vouros Inherently Interpretable Deep Reinforcement Learning through Online Mimicking Inproceedings EXTRAAMAS @AAMAS 2023, 2023. @inproceedings{Kontogiannis2023, title = {Inherently Interpretable Deep Reinforcement Learning through Online Mimicking}, author = {Andreas Kontogiannis, George Vouros}, year = {2023}, date = {2023-05-29}, booktitle = {EXTRAAMAS @AAMAS 2023}, abstract = {Although deep reinforcement learning (DRL) methods have been successfully applied in challenging tasks, their application in real-world operational settings – where transparency and accountability play important roles in automation – is challenged by methods’ limited ability to provide explanations. Among the paradigms for explainability in DRL is the interpretable box design paradigm, where interpretable models substitute inner closed constituent models of the DRL method, thus making the DRL method “inherently” interpretable. In this paper we propose a generic paradigm where interpretable policy models are trained following an online mimicking paradigm. We exemplify this paradigm through XDQN, an explainable variation of DQN that uses an interpretable policy model trained online with the deep policy model. XDQN is challenged in a complex, real-world operational multi-agent problem pertaining to the demand-capacity balancing problem of air traffic management (ATM), where human operators need to master complexity and understand the factors driving decision making. XDQN is shown to achieve performance similar to that of its DQN counterpart, while its abilities to provide global models’ interpretations and interpretations of local decisions are demonstrated.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Although deep reinforcement learning (DRL) methods have been successfully applied in challenging tasks, their application in real-world operational settings – where transparency and accountability play important roles in automation – is challenged by methods’ limited ability to provide explanations. Among the paradigms for explainability in DRL is the interpretable box design paradigm, where interpretable models substitute inner closed constituent models of the DRL method, thus making the DRL method “inherently" interpretable. In this paper we propose a generic paradigm where interpretable policy models are trained following an online mimicking paradigm. We exemplify this paradigm through XDQN, an explainable variation of DQN that uses an interpretable policy model trained online with the deep policy model. XDQN is challenged in a complex, real-world operational multi-agent problem pertaining to the demand-capacity balancing problem of air traffic management (ATM), where human operators need to master complexity and understand the factors driving decision making. XDQN is shown to achieve performance similar to that of its DQN counterpart, while its abilities to provide global models’ interpretations and interpretations of local decisions are demonstrated. |
Theocharis Kravaris Konstantinos Lentzos, Georgios Santipantakis George Vouros Gennady Andrienko Natalia Andrienko Ian Crook Jose Manuel Cordero Garcia Enrique Iglesias Martinez A L V Explaining deep reinforcement learning decisions in complex multiagent settings: towards enabling automation in air traffic flow management Journal Article Applied Intelligence, 53 , pp. 4063-4098, 2023. @article{Kravaris2023, title = {Explaining deep reinforcement learning decisions in complex multiagent settings: towards enabling automation in air traffic flow management}, author = {Theocharis Kravaris, Konstantinos Lentzos, Georgios Santipantakis, George A. Vouros, Gennady L. Andrienko, Natalia V. Andrienko, Ian Crook, Jose Manuel Cordero Garcia, Enrique Iglesias Martinez}, doi = {10.1007/s10489-022-03605-1}, year = {2023}, date = {2023-01-20}, journal = {Applied Intelligence}, volume = {53}, pages = {4063-4098}, abstract = {With the objective to enhance human performance and maximize engagement during the performance of tasks, we aim to advance automation for decision making in complex and large-scale multi-agent settings. Towards these goals, this paper presents a deep multi agent reinforcement learning method for resolving demand – capacity imbalances in real-world Air Traffic Management settings with thousands of agents. Agents comprising the system are able to jointly decide on the measures to be applied to resolve imbalances, while they provide explanations on their decisions: This information is rendered and explored via appropriate visual analytics tools. The paper presents how major challenges of scalability and complexity are addressed, and provides results from evaluation tests that show the abilities of models to provide high-quality solutions and high-fidelity explanations.}, keywords = {}, pubstate = {published}, tppubtype = {article} } With the objective to enhance human performance and maximize engagement during the performance of tasks, we aim to advance automation for decision making in complex and large-scale multi-agent settings. Towards these goals, this paper presents a deep multi agent reinforcement learning method for resolving demand – capacity imbalances in real-world Air Traffic Management settings with thousands of agents. Agents comprising the system are able to jointly decide on the measures to be applied to resolve imbalances, while they provide explanations on their decisions: This information is rendered and explored via appropriate visual analytics tools. The paper presents how major challenges of scalability and complexity are addressed, and provides results from evaluation tests that show the abilities of models to provide high-quality solutions and high-fidelity explanations. |
Christos Spatharis Alevizos Bastas, Theocharis Kravaris Konstantinos Blekas George Vouros Jose Manuel Cordero Garcia A Hierarchical multiagent reinforcement learning schemes for air traffic management Journal Article Neural Computing and Applications, 35 , pp. 147-159, 2023. @article{Spatharis2023, title = {Hierarchical multiagent reinforcement learning schemes for air traffic management}, author = {Christos Spatharis, Alevizos Bastas, Theocharis Kravaris, Konstantinos Blekas, George A. Vouros, Jose Manuel Cordero Garcia}, doi = {10.1007/s00521-021-05748-7}, year = {2023}, date = {2023-01-01}, journal = {Neural Computing and Applications}, volume = {35}, pages = {147-159}, abstract = {In this work we investigate the use of hierarchical multiagent reinforcement learning methods for the computation of policies to resolve congestion problems in the air traffic management domain. To address cases where the demand of airspace use exceeds capacity, we consider agents representing flights, who need to decide on ground delays at the pre-tactical stage of operations, towards executing their trajectories while adhering to airspace capacity constraints. Hierarchical reinforcement learning manages to handle real-world problems with high complexity, by partitioning the task into hierarchies of states and/or actions. This provides an efficient way of exploring the state–action space and constructing an advantageous decision-making mechanism. We first establish a general framework of hierarchical multiagent reinforcement learning, and then, we further formulate four alternative schemes of abstractions, on states, actions, or both. To quantitatively assess the quality of solutions of the proposed approaches and show the potential of the hierarchical methods in resolving the demand–capacity balance problem, we provide experimental results on real-world evaluation cases, where we measure the average delay per flight and the number of flights with delays.}, keywords = {}, pubstate = {published}, tppubtype = {article} } In this work we investigate the use of hierarchical multiagent reinforcement learning methods for the computation of policies to resolve congestion problems in the air traffic management domain. To address cases where the demand of airspace use exceeds capacity, we consider agents representing flights, who need to decide on ground delays at the pre-tactical stage of operations, towards executing their trajectories while adhering to airspace capacity constraints. Hierarchical reinforcement learning manages to handle real-world problems with high complexity, by partitioning the task into hierarchies of states and/or actions. This provides an efficient way of exploring the state–action space and constructing an advantageous decision-making mechanism. We first establish a general framework of hierarchical multiagent reinforcement learning, and then, we further formulate four alternative schemes of abstractions, on states, actions, or both. To quantitatively assess the quality of solutions of the proposed approaches and show the potential of the hierarchical methods in resolving the demand–capacity balance problem, we provide experimental results on real-world evaluation cases, where we measure the average delay per flight and the number of flights with delays. |
Vouros, George A Explainable Deep Reinforcement Learning: State of the Art and Challenges Journal Article ACM Computing Surveys, 55 (5), pp. 1-39, 2023. @article{Vouros2023, title = {Explainable Deep Reinforcement Learning: State of the Art and Challenges}, author = {George A. Vouros}, doi = {10.1145/3527448}, year = {2023}, date = {2023-01-01}, journal = {ACM Computing Surveys}, volume = {55}, number = {5}, pages = {1-39}, abstract = {Interpretability, explainability, and transparency are key issues to introducing artificial intelligence methods in many critical domains. This is important due to ethical concerns and trust issues strongly connected to reliability, robustness, auditability, and fairness, and has important consequences toward keeping the human in the loop in high levels of automation, especially in critical cases for decision making, where both (human and the machine) play important roles. Although the research community has given much attention to explainability of closed (or black) prediction boxes, there are tremendous needs for explainability of closed-box methods that support agents to act autonomously in the real world. Reinforcement learning methods, and especially their deep versions, are such closed-box methods. In this article, we aim to provide a review of state-of-the-art methods for explainable deep reinforcement learning methods, taking also into account the needs of human operators—that is, of those who make the actual and critical decisions in solving real-world problems. We provide a formal specification of the deep reinforcement learning explainability problems, and we identify the necessary components of a general explainable reinforcement learning framework. Based on these, we provide a comprehensive review of state-of-the-art methods, categorizing them into classes according to the paradigm they follow, the interpretable models they use, and the surface representation of explanations provided. The article concludes by identifying open questions and important challenges.}, keywords = {}, pubstate = {published}, tppubtype = {article} } Interpretability, explainability, and transparency are key issues to introducing artificial intelligence methods in many critical domains. This is important due to ethical concerns and trust issues strongly connected to reliability, robustness, auditability, and fairness, and has important consequences toward keeping the human in the loop in high levels of automation, especially in critical cases for decision making, where both (human and the machine) play important roles. Although the research community has given much attention to explainability of closed (or black) prediction boxes, there are tremendous needs for explainability of closed-box methods that support agents to act autonomously in the real world. Reinforcement learning methods, and especially their deep versions, are such closed-box methods. In this article, we aim to provide a review of state-of-the-art methods for explainable deep reinforcement learning methods, taking also into account the needs of human operators—that is, of those who make the actual and critical decisions in solving real-world problems. We provide a formal specification of the deep reinforcement learning explainability problems, and we identify the necessary components of a general explainable reinforcement learning framework. Based on these, we provide a comprehensive review of state-of-the-art methods, categorizing them into classes according to the paradigm they follow, the interpretable models they use, and the surface representation of explanations provided. The article concludes by identifying open questions and important challenges. |
2022 |
Natividad Valle M Florencia Lema, José Manuel Cordero Enrique Iglesias Rubén Rodríguez George Vouros Theocharis Kravaris George Papadopoulos Alevizos Bastas Georgios Santipantakis A Explainability & Transparency in higher levels of automation in the ATM domain Inproceedings SESAR Innovation Days 2022, 2022. @inproceedings{Valle2022, title = {Explainability & Transparency in higher levels of automation in the ATM domain}, author = {Natividad Valle, M Florencia Lema, José Manuel Cordero, Enrique Iglesias , Rubén Rodríguez, George A. Vouros, Theocharis Kravaris, George Papadopoulos, Alevizos Bastas, Georgios Santipantakis}, url = {https://www.sesarju.eu/sesarinnovationdays}, year = {2022}, date = {2022-12-06}, booktitle = {SESAR Innovation Days 2022}, journal = {IEEE Comput Graph Appl.}, abstract = {This paper presents findings, lessons learnt and guidelines for the use of explainable and transparent Artificial Intelligence (AI)/Machine Learning (ML) in ATM. The paper focuses on the results obtained from validating two AI/ML prototypes for Conflict Detection & Resolution (CD&R) and Air Traffic Flow Capacity Management (ATFCM) problems. These two prototypes are representative of the type of advanced automated systems that can support respectively the tactical and the pre-tactical operational phases The aim is, shifting the paradigm of human-AI teaming, providing full explainability and operational transparency. The major question is: when and how explanations should be provided for systems to be acceptable and trustworthy by operators?}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } This paper presents findings, lessons learnt and guidelines for the use of explainable and transparent Artificial Intelligence (AI)/Machine Learning (ML) in ATM. The paper focuses on the results obtained from validating two AI/ML prototypes for Conflict Detection & Resolution (CD&R) and Air Traffic Flow Capacity Management (ATFCM) problems. These two prototypes are representative of the type of advanced automated systems that can support respectively the tactical and the pre-tactical operational phases The aim is, shifting the paradigm of human-AI teaming, providing full explainability and operational transparency. The major question is: when and how explanations should be provided for systems to be acceptable and trustworthy by operators? |
George A. Vouros Theodore Tranos, Konstantinos Blekas Georgios Santipantakis Marc Melgosa Xavier Prats Data-driven estimation of flights’ hidden parameters Inproceedings SESAR Innovation Days 2022, 2022. @inproceedings{Vouros2022c, title = {Data-driven estimation of flights’ hidden parameters }, author = {George A. Vouros, Theodore Tranos, Konstantinos Blekas, Georgios Santipantakis, Marc Melgosa, Xavier Prats }, url = {https://www.sesarju.eu/sesarinnovationdays}, year = {2022}, date = {2022-12-06}, booktitle = {SESAR Innovation Days 2022}, abstract = {This paper presents a data-driven methodology for the estimation of flights’ hidden parameters, combining mechanistic and AI/ML models. In the context of this methodology the paper studies several AI/ML methods and reports on evaluation results for estimating hidden parameters, in terms of mean absolute error. In addition to the estimation of hidden parameters themselves, this paper examines how these estimations affect the prediction of KPIs regarding the efficiency of flights using a mechanistic model. Results show the accuracy of the proposed methods and the benefits of the proposed methodology. Indeed, the results show significant advances of data-driven methods to estimate hidden parameters towards predicting KPIs.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } This paper presents a data-driven methodology for the estimation of flights’ hidden parameters, combining mechanistic and AI/ML models. In the context of this methodology the paper studies several AI/ML methods and reports on evaluation results for estimating hidden parameters, in terms of mean absolute error. In addition to the estimation of hidden parameters themselves, this paper examines how these estimations affect the prediction of KPIs regarding the efficiency of flights using a mechanistic model. Results show the accuracy of the proposed methods and the benefits of the proposed methodology. Indeed, the results show significant advances of data-driven methods to estimate hidden parameters towards predicting KPIs. |
Alevizos Bastas, George Vouros A Data-driven prediction of Air Traffic Controllers reactions to resolving conflicts Journal Article Information Sciences, 613 , pp. 763-785, 2022. @article{Bastas2022, title = {Data-driven prediction of Air Traffic Controllers reactions to resolving conflicts}, author = {Alevizos Bastas, George A. Vouros}, doi = {10.1016/j.ins.2022.09.015}, year = {2022}, date = {2022-10-01}, journal = {Information Sciences}, volume = {613}, pages = {763-785}, abstract = {With the aim to enhance automation in conflict detection and resolution (CD&R) tasks in the Air Traffic Management domain, this article proposes deep learning (DL) techniques that model Air Traffic Controllers’ reactions in resolving conflicts violating aircraft trajectories separation minimum constraints: This implies learning when the Air Traffic Controller reacts towards resolving a conflict, and how he/she reacts. Timely reactions, to which this article aims, focus on when do reactions happen, aiming to predict the trajectory points, as the aircraft state evolves, that the Air Traffic Controller (ATCO) issues a conflict resolution action. Towards this goal, the article formulates the Air Traffic Controllers’ reaction prediction problem for CD&R, presents DL methods that can model Air Traffic Controllers’ timely reactions, and evaluates these methods in real-world data sets, showing their efficacy in solving the problem with very high accuracy.}, keywords = {}, pubstate = {published}, tppubtype = {article} } With the aim to enhance automation in conflict detection and resolution (CD&R) tasks in the Air Traffic Management domain, this article proposes deep learning (DL) techniques that model Air Traffic Controllers’ reactions in resolving conflicts violating aircraft trajectories separation minimum constraints: This implies learning when the Air Traffic Controller reacts towards resolving a conflict, and how he/she reacts. Timely reactions, to which this article aims, focus on when do reactions happen, aiming to predict the trajectory points, as the aircraft state evolves, that the Air Traffic Controller (ATCO) issues a conflict resolution action. Towards this goal, the article formulates the Air Traffic Controllers’ reaction prediction problem for CD&R, presents DL methods that can model Air Traffic Controllers’ timely reactions, and evaluates these methods in real-world data sets, showing their efficacy in solving the problem with very high accuracy. |
Vouros, George A Tutorial on Explainable Deep Reinforcement Learning: One framework, three paradigms and many challenges Inproceedings SETN 2022, ACM, 2022. @inproceedings{Vouros2022, title = {Tutorial on Explainable Deep Reinforcement Learning: One framework, three paradigms and many challenges}, author = {George A. Vouros}, doi = {10.1145/3549737.3549808}, year = {2022}, date = {2022-09-10}, booktitle = {SETN 2022}, publisher = {ACM}, abstract = {Interpretability, explainability and transparency are key issues to introducing Artificial Intelligence closed—box methods in many critical domains: This is important due to ethical concerns and trust issues strongly connected to reliability, robustness, auditability and fairness, and has important consequences towards keeping the human in the loop in high levels of automation, especially in critical cases for decision making. Reinforcement learning methods, and especially their deep versions, are closed-box methods that support agents to act autonomously in the real world. This tutorial will provide a formal specification of the deep reinforcement learning explainability problems, and will present the necessary components of a general explainable reinforcement learning framework. Based on this framework will provide distinct explainability paradigms towards solving explainability problems, with examples from state-of-the-art methods and real-world cases. The tutorial will conclude identifying open questions and important challenges. The tutorial is based on the survey paper on “Explainable Deep Reinforcement Learning” State of the Art and Challenges”}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Interpretability, explainability and transparency are key issues to introducing Artificial Intelligence closed—box methods in many critical domains: This is important due to ethical concerns and trust issues strongly connected to reliability, robustness, auditability and fairness, and has important consequences towards keeping the human in the loop in high levels of automation, especially in critical cases for decision making. Reinforcement learning methods, and especially their deep versions, are closed-box methods that support agents to act autonomously in the real world. This tutorial will provide a formal specification of the deep reinforcement learning explainability problems, and will present the necessary components of a general explainable reinforcement learning framework. Based on this framework will provide distinct explainability paradigms towards solving explainability problems, with examples from state-of-the-art methods and real-world cases. The tutorial will conclude identifying open questions and important challenges. The tutorial is based on the survey paper on “Explainable Deep Reinforcement Learning” State of the Art and Challenges" |
Apostolos Glenis, George Vouros A SCALE-BOSS: A framework for scalable time-series classification using symbolic representations Inproceedings SETN 2022, ACM, 2022. @inproceedings{Glenis2022, title = {SCALE-BOSS: A framework for scalable time-series classification using symbolic representations}, author = {Apostolos Glenis, George A. Vouros}, doi = {10.1145/3549737.3549761}, year = {2022}, date = {2022-09-10}, booktitle = {SETN 2022}, publisher = {ACM}, abstract = {Time-Series Classification (TSC) is an important problem in many fields across sciences. Many algorithms for TSC use symbolic representation to combat noise. In this paper we propose a framework, namely SCALE-BOSS, to build TSC algorithms that exploit time-series models based on symbolic representations. While alternative symbolic representations can be incorporated, we have opted to use the Bag-Of-SFA (BOSS) approach, and thus SFA, as a state-of-the-art symbolic time series representation. We investigate the efficiency of several instantiations of this framework based on two main variations, where the TSC model is built either by a time-series classification or by a clustering algorithm. The objective is to advance the computational efficiency of TSC classification algorithms without sacrificing their accuracy. We evaluate the instantiations of the SCALE-BOSS framework on those datasets in the UCR time-series repository that include the largest training sets. Comparisons with state of the art methods on TSC show the balance between computational efficiency and accuracy on predictions achieved.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Time-Series Classification (TSC) is an important problem in many fields across sciences. Many algorithms for TSC use symbolic representation to combat noise. In this paper we propose a framework, namely SCALE-BOSS, to build TSC algorithms that exploit time-series models based on symbolic representations. While alternative symbolic representations can be incorporated, we have opted to use the Bag-Of-SFA (BOSS) approach, and thus SFA, as a state-of-the-art symbolic time series representation. We investigate the efficiency of several instantiations of this framework based on two main variations, where the TSC model is built either by a time-series classification or by a clustering algorithm. The objective is to advance the computational efficiency of TSC classification algorithms without sacrificing their accuracy. We evaluate the instantiations of the SCALE-BOSS framework on those datasets in the UCR time-series repository that include the largest training sets. Comparisons with state of the art methods on TSC show the balance between computational efficiency and accuracy on predictions achieved. |
Georgios M. Santipantakis Konstantinos I. Kotis, Apostolos Glenis George Vouros Christos Doulkeridis Akrivi Vlachou A RDF-Gen: generating RDF triples from big data sources Journal Article Knowledge and Information Systems, 64 , pp. 2985–3015, 2022. @article{Santipantakis2022, title = {RDF-Gen: generating RDF triples from big data sources}, author = {Georgios M. Santipantakis, Konstantinos I. Kotis, Apostolos Glenis, George A. Vouros, Christos Doulkeridis, Akrivi Vlachou}, doi = {10.1007/s10115-022-01729-x}, year = {2022}, date = {2022-08-13}, journal = {Knowledge and Information Systems}, volume = {64}, pages = {2985–3015}, abstract = {Transforming disparate and heterogeneous data sources that provide large volumes of data in high velocity into a common form allows integrated and enriched views on data and thus provides further opportunities to advance the effectiveness and accuracy of data analysis and prediction tasks. This paper presents the RDF-Gen approach for transforming data provided by archival and streaming data sources, provided in various formats, into RDF triples, according to a set of ontological specifications. RDF-Gen introduces a generic mechanism which supports the transformation of data efficiently (i.e., with high throughput and low latency), even in cases where the velocity of data presents high peaks, offering facilities for discovering associations between data from different sources, and supporting transformation of modular data sets. This paper presents a parallel implementation of RDF-Gen, also presenting data transformation workflows that allow variations incorporating RDF-Gen instances, adjusting to the needs of data sources, application areas and performance requirements. RDF-Gen is experimentally evaluated against state of the art, in both archival and streaming settings: Experimental results show RDF-Gen efficiency and highlight key contributions.}, keywords = {}, pubstate = {published}, tppubtype = {article} } Transforming disparate and heterogeneous data sources that provide large volumes of data in high velocity into a common form allows integrated and enriched views on data and thus provides further opportunities to advance the effectiveness and accuracy of data analysis and prediction tasks. This paper presents the RDF-Gen approach for transforming data provided by archival and streaming data sources, provided in various formats, into RDF triples, according to a set of ontological specifications. RDF-Gen introduces a generic mechanism which supports the transformation of data efficiently (i.e., with high throughput and low latency), even in cases where the velocity of data presents high peaks, offering facilities for discovering associations between data from different sources, and supporting transformation of modular data sets. This paper presents a parallel implementation of RDF-Gen, also presenting data transformation workflows that allow variations incorporating RDF-Gen instances, adjusting to the needs of data sources, application areas and performance requirements. RDF-Gen is experimentally evaluated against state of the art, in both archival and streaming settings: Experimental results show RDF-Gen efficiency and highlight key contributions. |
George A. Vouros George Papadopoulos, Alevizos Bastas Jose Manuel Cordero Garcia Ruben Rodrigez Rodrigez Automating the Resolution of Flight Conflicts: Deep Reinforcement Learning in Service of Air Traffic Controllers Inproceedings PAIS 2022, IOS PRess, 2022. @inproceedings{Vouros2022b, title = {Automating the Resolution of Flight Conflicts: Deep Reinforcement Learning in Service of Air Traffic Controllers}, author = {George A. Vouros, George Papadopoulos, Alevizos Bastas, Jose Manuel Cordero Garcia, Ruben Rodrigez Rodrigez}, doi = {10.3233/FAIA220066}, year = {2022}, date = {2022-07-01}, booktitle = {PAIS 2022}, volume = {351}, publisher = {IOS PRess}, series = {Frontiers in Artificial Intelligence and Applications}, abstract = {Dense and complex air traffic scenarios require higher levels of automation than those exhibited by tactical conflict detection and resolution (CD&R) tools that air traffic controllers (ATCO) use today. However, the air traffic control (ATC) domain, being safety critical, requires AI systems to which operators are comfortable to relinquishing control, guaranteeing operational integrity and automation adoption. Two major factors towards this goal are quality of solutions, and transparency in decision making. This paper proposes using a graph convolutional reinforcement learning method operating in a multiagent setting where each agent (flight) performs a CD&R task, jointly with other agents. We show that this method can provide high-quality solutions with respect to stakeholders interests (air traffic controllers and airspace users), addressing operational transparency issues.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Dense and complex air traffic scenarios require higher levels of automation than those exhibited by tactical conflict detection and resolution (CD&R) tools that air traffic controllers (ATCO) use today. However, the air traffic control (ATC) domain, being safety critical, requires AI systems to which operators are comfortable to relinquishing control, guaranteeing operational integrity and automation adoption. Two major factors towards this goal are quality of solutions, and transparency in decision making. This paper proposes using a graph convolutional reinforcement learning method operating in a multiagent setting where each agent (flight) performs a CD&R task, jointly with other agents. We show that this method can provide high-quality solutions with respect to stakeholders interests (air traffic controllers and airspace users), addressing operational transparency issues. |
Gennady L. Andrienko Natalia V. Andrienko, Jose Manuel Cordero Garcia Dirk Hecker George Vouros A Supporting Visual Exploration of Iterative Job Scheduling Journal Article IEEE Comput Graph Appl . , 42 (3), pp. 74-86, 2022. @article{Andrienko2022, title = {Supporting Visual Exploration of Iterative Job Scheduling}, author = {Gennady L. Andrienko, Natalia V. Andrienko, Jose Manuel Cordero Garcia, Dirk Hecker, George A. Vouros}, doi = {10.1109/MCG.2022.3163437}, year = {2022}, date = {2022-05-06}, journal = {IEEE Comput Graph Appl . }, volume = {42}, number = {3}, pages = {74-86}, abstract = {We consider the general problem known as job shop scheduling, in which multiple jobs consist of sequential operations that need to be executed or served by appropriate machines having limited capacities. For example, train journeys (jobs) consist of moves and stops (operations) to be served by rail tracks and stations (machines). A schedule is an assignment of the job operations to machines and times where and when they will be executed. The developers of computational methods for job scheduling need tools enabling them to explore how their methods work. At a high level of generality, we define the system of pertinent exploration tasks and a combination of visualizations capable of supporting the tasks. We provide general descriptions of the purposes, contents, visual encoding, properties, and interactive facilities of the visualizations and illustrate them with images from an example implementation in air traffic management. We justify the design of the visualizations based on the tasks, principles of creating visualizations for pattern discovery, and scalability requirements. The outcomes of our research are sufficiently general to be of use in a variety of applications.}, keywords = {}, pubstate = {published}, tppubtype = {article} } We consider the general problem known as job shop scheduling, in which multiple jobs consist of sequential operations that need to be executed or served by appropriate machines having limited capacities. For example, train journeys (jobs) consist of moves and stops (operations) to be served by rail tracks and stations (machines). A schedule is an assignment of the job operations to machines and times where and when they will be executed. The developers of computational methods for job scheduling need tools enabling them to explore how their methods work. At a high level of generality, we define the system of pertinent exploration tasks and a combination of visualizations capable of supporting the tasks. We provide general descriptions of the purposes, contents, visual encoding, properties, and interactive facilities of the visualizations and illustrate them with images from an example implementation in air traffic management. We justify the design of the visualizations based on the tasks, principles of creating visualizations for pattern discovery, and scalability requirements. The outcomes of our research are sufficiently general to be of use in a variety of applications. |
2021 |
Nikolaos Koutroumanis Georgios M. Santipantakis, Apostolos Glenis Christos Doulkeridis & George Vouros A Scalable enrichment of mobility data with weather information Journal Article GEOINFORMATICA, 25 , pp. 291-309, 2021. @article{Koutroumanis2021, title = {Scalable enrichment of mobility data with weather information}, author = {Nikolaos Koutroumanis, Georgios M. Santipantakis, Apostolos Glenis, Christos Doulkeridis & George A. Vouros }, doi = {10.1007/s10707-020-00423-w}, year = {2021}, date = {2021-09-17}, journal = {GEOINFORMATICA}, volume = {25}, pages = {291-309}, abstract = {More and more real-life applications for mobility analytics require the joint exploitation of positional information of moving objects together with weather data that correspond to the movement. In particular, this is evident in fleet management applications for improved routing and reduced fuel consumption, in the maritime domain for more accurate trajectory prediction, as well as in air-traffic management for predicting regulations and reducing delays. Motivated by such applications, in this paper, we present a system for the enrichment of mobility data with weather information. Our main application scenario concerns streaming positional information (such as GPS traces of vehicles) that is collected and is enriched in an online fashion with stored weather data. We present the system architecture of a centralized version that runs on a single machine and exploits caching to improve its efficiency. Also, we extend our approach to a parallel implementation on top of Apache Kafka, which can scale to hundreds of thousands of processed records when provided with more computing nodes. Furthermore, we present extensions of our system for: (a) enrichment of more complex geometries than point data, and (b) providing linked RDF data as output. Our experimental evaluation on a medium-sized cluster shows the scalability of our approach in terms of number of processed records per second.}, keywords = {}, pubstate = {published}, tppubtype = {article} } More and more real-life applications for mobility analytics require the joint exploitation of positional information of moving objects together with weather data that correspond to the movement. In particular, this is evident in fleet management applications for improved routing and reduced fuel consumption, in the maritime domain for more accurate trajectory prediction, as well as in air-traffic management for predicting regulations and reducing delays. Motivated by such applications, in this paper, we present a system for the enrichment of mobility data with weather information. Our main application scenario concerns streaming positional information (such as GPS traces of vehicles) that is collected and is enriched in an online fashion with stored weather data. We present the system architecture of a centralized version that runs on a single machine and exploits caching to improve its efficiency. Also, we extend our approach to a parallel implementation on top of Apache Kafka, which can scale to hundreds of thousands of processed records when provided with more computing nodes. Furthermore, we present extensions of our system for: (a) enrichment of more complex geometries than point data, and (b) providing linked RDF data as output. Our experimental evaluation on a medium-sized cluster shows the scalability of our approach in terms of number of processed records per second. |
Panagiotis Nikitopoulos Akrivi Vlachou, Christos Doulkeridis George Vouros A Parallel and scalable processing of spatio-temporal RDF queries using Spark Journal Article Geoinformatica, 25 , pp. 623-653, 2021. @article{Nikitopoulos2021, title = {Parallel and scalable processing of spatio-temporal RDF queries using Spark}, author = {Panagiotis Nikitopoulos, Akrivi Vlachou, Christos Doulkeridis, George A. Vouros}, doi = {10.1007/s10707-019-00371-0}, year = {2021}, date = {2021-07-03}, journal = {Geoinformatica}, volume = {25}, pages = {623-653}, abstract = {The ever-increasing size of data emanating from mobile devices and sensors, dictates the use of distributed systems for storing and querying these data. Typically, such data sources provide some spatio-temporal information, alongside other useful data. The RDF data model can be used to interlink and exchange data originating from heterogeneous sources in a uniform manner. For example, consider the case where vessels report their spatio-temporal position, on a regular basis, by using various surveillance systems. In this scenario, a user might be interested to know which vessels were moving in a specific area for a given temporal range. In this paper, we address the problem of efficiently storing and querying spatio-temporal RDF data in parallel. We specifically study the case of SPARQL queries with spatio-temporal constraints, by proposing the DiStRDF system, which is comprised of a Storage and a Processing Layer. The DiStRDF Storage Layer is responsible for efficiently storing large amount of historical spatio-temporal RDF data of moving objects. On top of it, we devise our DiStRDF Processing Layer, which parses a SPARQL query and produces corresponding logical and physical execution plans. We use Spark, a well-known distributed in-memory processing framework, as the underlying processing engine. Our experimental evaluation, on real data from both aviation and maritime domains, demonstrates the efficiency of our DiStRDF system, when using various spatio-temporal range constraints.}, keywords = {}, pubstate = {published}, tppubtype = {article} } The ever-increasing size of data emanating from mobile devices and sensors, dictates the use of distributed systems for storing and querying these data. Typically, such data sources provide some spatio-temporal information, alongside other useful data. The RDF data model can be used to interlink and exchange data originating from heterogeneous sources in a uniform manner. For example, consider the case where vessels report their spatio-temporal position, on a regular basis, by using various surveillance systems. In this scenario, a user might be interested to know which vessels were moving in a specific area for a given temporal range. In this paper, we address the problem of efficiently storing and querying spatio-temporal RDF data in parallel. We specifically study the case of SPARQL queries with spatio-temporal constraints, by proposing the DiStRDF system, which is comprised of a Storage and a Processing Layer. The DiStRDF Storage Layer is responsible for efficiently storing large amount of historical spatio-temporal RDF data of moving objects. On top of it, we devise our DiStRDF Processing Layer, which parses a SPARQL query and produces corresponding logical and physical execution plans. We use Spark, a well-known distributed in-memory processing framework, as the underlying processing engine. Our experimental evaluation, on real data from both aviation and maritime domains, demonstrates the efficiency of our DiStRDF system, when using various spatio-temporal range constraints. |
Georgios M. Santipantakis George A. Vouros, Christos Doulkeridis Coronis: Towards Integrated and Open COVID-19 Data Conference EDBT 2021, 2021. @conference{Santipantakis2021, title = {Coronis: Towards Integrated and Open COVID-19 Data}, author = {Georgios M. Santipantakis, George A. Vouros, Christos Doulkeridis}, year = {2021}, date = {2021-03-26}, booktitle = {EDBT 2021}, keywords = {}, pubstate = {published}, tppubtype = {conference} } |
Georgios M. Santipantakis Christos Doulkeridis, George Vouros A Link Discovery for Maritime Monitoring Book Chapter Springer, 2021, ISBN: 978-3-030-61852-0. @inbook{Santipantakis2021b, title = {Link Discovery for Maritime Monitoring}, author = {Georgios M. Santipantakis, Christos Doulkeridis, George A. Vouros}, isbn = {978-3-030-61852-0}, year = {2021}, date = {2021-02-09}, publisher = {Springer}, abstract = {Link discovery in the maritime domain is the process of identifying relations—usually of spatial or spatio-temporal nature—between entities that originate from different data sources. Essentially, link discovery is a step towards data integration, which enables interlinking data from disparate sources. As a typical example, vessel trajectories need to be enriched with various types of information: weather conditions, events, contextual data. In turn, this provides enriched data descriptions to data analysis operations, which may lead to the identification of hidden or complex patterns, which would otherwise not be discovered, as they rely on data originating from disparate data sources. This chapter presents the fundamental concepts of link discovery relevant to the maritime domain, focusing on spatial and spatio-temporal data. Due to the processing-intensive nature of the link discovery task over voluminous data, several techniques for efficient processing are presented together with examples on real-world data from the maritime domain.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Link discovery in the maritime domain is the process of identifying relations—usually of spatial or spatio-temporal nature—between entities that originate from different data sources. Essentially, link discovery is a step towards data integration, which enables interlinking data from disparate sources. As a typical example, vessel trajectories need to be enriched with various types of information: weather conditions, events, contextual data. In turn, this provides enriched data descriptions to data analysis operations, which may lead to the identification of hidden or complex patterns, which would otherwise not be discovered, as they rely on data originating from disparate data sources. This chapter presents the fundamental concepts of link discovery relevant to the maritime domain, focusing on spatial and spatio-temporal data. Due to the processing-intensive nature of the link discovery task over voluminous data, several techniques for efficient processing are presented together with examples on real-world data from the maritime domain. |
2020 |
Apostolos Glenis, George Vouros A Balancing Between Scalability and Accuracy in Time-Series Classification for Stream and Batch Settings Conference International Conference on Discovery Science, Lecture Notes in Computer Science, 12323 , LNAI Springer, 2020. @conference{Glenis2020, title = {Balancing Between Scalability and Accuracy in Time-Series Classification for Stream and Batch Settings}, author = {Apostolos Glenis, George A. Vouros}, year = {2020}, date = {2020-10-15}, booktitle = {International Conference on Discovery Science, Lecture Notes in Computer Science}, volume = {12323}, pages = {265-279}, publisher = {Springer}, series = {LNAI}, abstract = {As big data sources providing time series increase, and data is provided in increased velocity and volume, we need to efficiently recognize data provided, classifying it according to their type, origin etc. This is a first important step in doing analytics on data provided from disparate data sources, such as archival sources, multiple sensors, or social media feeds. Time series classification is the task labeling time series using a set of predefined labels. In this paper we present the K-BOSS-VS algorithm for time series classification. The proposed algorithm is based on state-of-the-art symbolic time series classification algorithms, and aims to achieve high accuracy, balancing with computational efficiency. K-BOSS-VS exploits K representatives of each time series class to classify new series. This provides opportunities for representing intra-class differences, thus increasing the classification accuracy, while incurring a small performance overhead compared to methods using one class representative. Additionally, K-BOSS-VS offers a solution for classifying time-series in batch and streaming settings, due to the opportunities for increasing computational efficiency and the low memory requirements.}, keywords = {}, pubstate = {published}, tppubtype = {conference} } As big data sources providing time series increase, and data is provided in increased velocity and volume, we need to efficiently recognize data provided, classifying it according to their type, origin etc. This is a first important step in doing analytics on data provided from disparate data sources, such as archival sources, multiple sensors, or social media feeds. Time series classification is the task labeling time series using a set of predefined labels. In this paper we present the K-BOSS-VS algorithm for time series classification. The proposed algorithm is based on state-of-the-art symbolic time series classification algorithms, and aims to achieve high accuracy, balancing with computational efficiency. K-BOSS-VS exploits K representatives of each time series class to classify new series. This provides opportunities for representing intra-class differences, thus increasing the classification accuracy, while incurring a small performance overhead compared to methods using one class representative. Additionally, K-BOSS-VS offers a solution for classifying time-series in batch and streaming settings, due to the opportunities for increasing computational efficiency and the low memory requirements. |
Christos Spatharis Konstantinos Blekas, George Vouros A Apprenticeship learning of flight trajectories prediction with inverse reinforcement learning. Inproceedings SETN 2020: 11th Hellenic Conference on Artificial Intelligence, Athens, Greece, ACM, 2020. @inproceedings{Spatharis2020, title = {Apprenticeship learning of flight trajectories prediction with inverse reinforcement learning.}, author = {Christos Spatharis, Konstantinos Blekas, George A. Vouros}, doi = {10.1145/3411408.3411427}, year = {2020}, date = {2020-09-01}, booktitle = {SETN 2020: 11th Hellenic Conference on Artificial Intelligence, Athens, Greece}, publisher = {ACM}, abstract = {One of the primary goals of Artificial Intelligence research is to develop machines with human-like intelligence, perception and reasoning. In this direction teaching apprentice agents by observing demonstrations delivered by experts is a framework of imitation learning that can provide improved solutions and it is possible to significantly outperform the demonstrator. Inverse reinforcement learning (IRL) is a paradigm relying on Markov Decision Processes (MDPs) that has a twofold target: to learn optimum policies of autonomous agents for solving complex tasks from successful demonstrations, and also to discover the unknown reward function that could explain the expert behavior. In this article we are addressing the trajectory prediction problem in the aviation domain by using an IRL approach. The proposed learning scheme provides an imitation process where the algorithm tries to imitate demonstrated trajectories, exploiting raw trajectory data enriched with contextual features and learn an efficient reward model that is learned during imitation and has generalization capabilities to unknown cases. We show several experimental results using real trajectory data from the Spanish FIR that confirms the effectiveness of our approach in automatically predicting trajectories.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } One of the primary goals of Artificial Intelligence research is to develop machines with human-like intelligence, perception and reasoning. In this direction teaching apprentice agents by observing demonstrations delivered by experts is a framework of imitation learning that can provide improved solutions and it is possible to significantly outperform the demonstrator. Inverse reinforcement learning (IRL) is a paradigm relying on Markov Decision Processes (MDPs) that has a twofold target: to learn optimum policies of autonomous agents for solving complex tasks from successful demonstrations, and also to discover the unknown reward function that could explain the expert behavior. In this article we are addressing the trajectory prediction problem in the aviation domain by using an IRL approach. The proposed learning scheme provides an imitation process where the algorithm tries to imitate demonstrated trajectories, exploiting raw trajectory data enriched with contextual features and learn an efficient reward model that is learned during imitation and has generalization capabilities to unknown cases. We show several experimental results using real trajectory data from the Spanish FIR that confirms the effectiveness of our approach in automatically predicting trajectories. |
Georgios M. Santipantakis Apostolos Glenis, Kostas Patroumpas Akrivi Vlachou Christos Doulkeridis George Vouros Nikos Pelekis Yannis Theodoridis A SPARTAN: Semantic integration of big spatio-temporal data from streaming and archival sources Journal Article Future Gener. Comput. Syst., 110 , pp. 540-555, 2020. @article{Santipantakis2020, title = {SPARTAN: Semantic integration of big spatio-temporal data from streaming and archival sources}, author = {Georgios M. Santipantakis, Apostolos Glenis, Kostas Patroumpas, Akrivi Vlachou, Christos Doulkeridis, George A. Vouros, Nikos Pelekis, Yannis Theodoridis}, doi = {10.1016/j.future.2018.07.007}, year = {2020}, date = {2020-06-13}, journal = {Future Gener. Comput. Syst.}, volume = {110}, pages = {540-555}, abstract = {An ever-increasing number of applications in critical domains, such as maritime and aviation, generate, collect, manage and process spatio-temporal data related to the mobility of entities. This wealth of data can be exploited for various purposes, towards improving the safety of operations, reducing economical costs, and increasing dependability: The major issue to achieve these objectives is increasing predictability of moving objects’ trajectories and events. To achieve this purpose in a data-driven way we need to exploit in integrated manners data from a variety of disparate and heterogeneous data sources, both streaming and archival, regarding – among other – surveillance, weather, and contextual data. Motivated by this fact, in this paper, we propose a framework for semantic integration of big mobility data with other data sources that are necessary to data analytics tasks, providing a unified representation of such data. Notable features of our framework include the real-time generation of data synopses of moving entities’ trajectories, the efficient and flexible transformation of data from heterogeneous and big data sources in RDF, and the spatio-temporal link discovery between spatio-temporal entities in diverse data sources. The design and implementation of our framework uses big data technologies (Apache Flink and Kafka), and our experimental evaluation demonstrates the efficiency and scalability of the proposed framework using large, real-life datasets.}, keywords = {}, pubstate = {published}, tppubtype = {article} } An ever-increasing number of applications in critical domains, such as maritime and aviation, generate, collect, manage and process spatio-temporal data related to the mobility of entities. This wealth of data can be exploited for various purposes, towards improving the safety of operations, reducing economical costs, and increasing dependability: The major issue to achieve these objectives is increasing predictability of moving objects’ trajectories and events. To achieve this purpose in a data-driven way we need to exploit in integrated manners data from a variety of disparate and heterogeneous data sources, both streaming and archival, regarding – among other – surveillance, weather, and contextual data. Motivated by this fact, in this paper, we propose a framework for semantic integration of big mobility data with other data sources that are necessary to data analytics tasks, providing a unified representation of such data. Notable features of our framework include the real-time generation of data synopses of moving entities’ trajectories, the efficient and flexible transformation of data from heterogeneous and big data sources in RDF, and the spatio-temporal link discovery between spatio-temporal entities in diverse data sources. The design and implementation of our framework uses big data technologies (Apache Flink and Kafka), and our experimental evaluation demonstrates the efficiency and scalability of the proposed framework using large, real-life datasets. |
Costas Vassilakis Konstantinos Kotis, Dimitris Spiliotopoulos Dionisis Margaris Vlasios Kasapakis Christos-Nikolaos Anagnostopoulos Georgios Santipantakis George Vouros Theodore Kotsilieris Volha Petukhova Andrei Malchanau Ioanna Lykourentzou Kaj Michael Helin Artem Revenko Nenad Gligoric Boris Pokric M A A Semantic Mixed Reality Framework for Shared Cultural Experiences Ecosystems Journal Article Big Data Cogn. Comput, 4 (2), 2020. @article{DBLP:journals/bdcc/VassilakisKSMKA20, title = {A Semantic Mixed Reality Framework for Shared Cultural Experiences Ecosystems}, author = {Costas Vassilakis, Konstantinos Kotis, Dimitris Spiliotopoulos, Dionisis Margaris, Vlasios Kasapakis, Christos-Nikolaos Anagnostopoulos, Georgios M. Santipantakis, George A. Vouros, Theodore Kotsilieris, Volha Petukhova, Andrei Malchanau, Ioanna Lykourentzou, Kaj Michael Helin, Artem Revenko, Nenad Gligoric, Boris Pokric}, doi = {10.3390/bdcc4020006}, year = {2020}, date = {2020-04-20}, journal = {Big Data Cogn. Comput}, volume = {4}, number = {2}, abstract = {This paper presents SemMR, a semantic framework for modelling interactions between human and non-human entities and managing reusable and optimized cultural experiences, towards a shared cultural experience ecosystem that might seamlessly accommodate mixed reality experiences. The SemMR framework synthesizes and integrates interaction data into semantically rich reusable structures and facilitates the interaction between different types of entities in a symbiotic way, within a large, virtual, and fully experiential open world, promoting experience sharing at the user level, as well as data/application interoperability and low-effort implementation at the software engineering level. The proposed semantic framework introduces methods for low-effort implementation and the deployment of open and reusable cultural content, applications, and tools, around the concept of cultural experience as a semantic trajectory or simply, experience as a trajectory (eX-trajectory). The methods facilitate the collection and analysis of data regarding the behaviour of users and their interaction with other users and the environment, towards optimizing eX-trajectories via reconfiguration. The SemMR framework supports the synthesis, enhancement, and recommendation of highly complex reconfigurable eX-trajectories, while using semantically integrated disparate and heterogeneous related data. Overall, this work aims to semantically manage interactions and experiences through the eX-trajectory concept, towards delivering enriched cultural experiences.}, keywords = {}, pubstate = {published}, tppubtype = {article} } This paper presents SemMR, a semantic framework for modelling interactions between human and non-human entities and managing reusable and optimized cultural experiences, towards a shared cultural experience ecosystem that might seamlessly accommodate mixed reality experiences. The SemMR framework synthesizes and integrates interaction data into semantically rich reusable structures and facilitates the interaction between different types of entities in a symbiotic way, within a large, virtual, and fully experiential open world, promoting experience sharing at the user level, as well as data/application interoperability and low-effort implementation at the software engineering level. The proposed semantic framework introduces methods for low-effort implementation and the deployment of open and reusable cultural content, applications, and tools, around the concept of cultural experience as a semantic trajectory or simply, experience as a trajectory (eX-trajectory). The methods facilitate the collection and analysis of data regarding the behaviour of users and their interaction with other users and the environment, towards optimizing eX-trajectories via reconfiguration. The SemMR framework supports the synthesis, enhancement, and recommendation of highly complex reconfigurable eX-trajectories, while using semantically integrated disparate and heterogeneous related data. Overall, this work aims to semantically manage interactions and experiences through the eX-trajectory concept, towards delivering enriched cultural experiences. |
George A. Vouros Gennady L. Andrienko, Christos Doulkeridis Nikolaos Pelekis Alexander Artikis Anne-Laure Jousselme Cyril Ray José Manuel Cordero Garcia David Scarlatti Big Data Analytics for Time-Critical Mobility Forecasting, From Raw Data to Trajectory-Oriented Mobility Analytics in the Aviation and Maritime Domains Book Springer, 2020, ISBN: 978-3-030-45163-9. @book{Vouros2020, title = {Big Data Analytics for Time-Critical Mobility Forecasting, From Raw Data to Trajectory-Oriented Mobility Analytics in the Aviation and Maritime Domains}, author = {George A. Vouros, Gennady L. Andrienko, Christos Doulkeridis, Nikolaos Pelekis, Alexander Artikis, Anne-Laure Jousselme, Cyril Ray, José Manuel Cordero Garcia, David Scarlatti}, isbn = {978-3-030-45163-9}, year = {2020}, date = {2020-03-01}, publisher = {Springer}, keywords = {}, pubstate = {published}, tppubtype = {book} } |
Karampelas, Andreas; Vouros, George A Time and Space Efficient Large Scale Link Discovery using String Similarities Journal Article Fundamenta Informaticae, 172 , pp. 299-325, 2020, ISSN: 0169-2968 (P). @article{351, title = {Time and Space Efficient Large Scale Link Discovery using String Similarities}, author = {Andreas Karampelas and George A Vouros}, url = {https://content.iospress.com/articles/fundamenta-informaticae/fi1906?resultNumber=0&totalResults=158&start=0&q=Time+and+Space+Efficient&resultsPageSize=10&rows=10}, doi = {10.3233/FI-2020-1906}, issn = {0169-2968 (P)}, year = {2020}, date = {2020-02-01}, journal = {Fundamenta Informaticae}, volume = {172}, pages = {299-325}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
Kotis, Konstantinos; Vouros, George A; Spiliotopoulos, Dimitris Ontology engineering methodologies for the evolution of living and reused ontologies: status, trends, findings and recommendations Journal Article The Knowledge Engineering Review, 35 , 2020. @article{358, title = {Ontology engineering methodologies for the evolution of living and reused ontologies: status, trends, findings and recommendations}, author = {Konstantinos Kotis and George A Vouros and Dimitris Spiliotopoulos}, url = {https://doi.org/10.1017/S0269888920000065}, doi = {https://doi.org/10.1017/S0269888920000065}, year = {2020}, date = {2020-01-01}, journal = {The Knowledge Engineering Review}, volume = {35}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
Vouros, George A; Glenis, Apostolis; Doulkeridis, Christos The delta big data architecture for mobility analytics Conference 2020 IEEE Sixth International Conference on Big Data Computing Service and Applications (BigDataService), IEEE IEEE, Oxford, UK, 2020. @conference{359, title = {The delta big data architecture for mobility analytics}, author = {George A Vouros and Apostolis Glenis and Christos Doulkeridis}, year = {2020}, date = {2020-01-01}, booktitle = {2020 IEEE Sixth International Conference on Big Data Computing Service and Applications (BigDataService)}, publisher = {IEEE}, address = {Oxford, UK}, organization = {IEEE}, keywords = {}, pubstate = {published}, tppubtype = {conference} } |
2019 |
Kravaris, Theocharis; et al., Resolving Congestions in the Air Traffic Management Domain via Multiagent Reinforcement Learning Methods Journal Article arXiv, cs.MA , 2019. @article{354, title = {Resolving Congestions in the Air Traffic Management Domain via Multiagent Reinforcement Learning Methods}, author = {Theocharis Kravaris and et al.}, url = {https://arxiv.org/pdf/1912.06860.pdf}, year = {2019}, date = {2019-12-01}, journal = {arXiv}, volume = {cs.MA}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
G.Vouros, ; Santipantakis, G; Doulkeridis, C; Vlachou, A; Andrienko, G; Andrienko, N; Fuchs, G; Martinez, Miguel Garcia; Cordero, Jose Manuel Garcia Journal Of Data Semantics, 8 , 2019. @article{352, title = {The datAcron Ontology for the Specification of Semantic Trajectories: Specification of Semantic Trajectories for Data Transformations Supporting Visual Analytics}, author = {G.Vouros and G Santipantakis and C Doulkeridis and A Vlachou and G Andrienko and N Andrienko and G Fuchs and Miguel Garcia Martinez and Jose Manuel Garcia Cordero}, url = {http://link.springer.com/article/10.1007/s13740-019-00108-0}, doi = {10.1007/s13740-019-00108-0}, year = {2019}, date = {2019-11-01}, journal = {Journal Of Data Semantics}, volume = {8}, chapter = {235}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
Petrou, P; et al., ARGO: A Big Data Framework for Online Trajectory Prediction Conference SSTD 2019, 2019. @conference{345, title = {ARGO: A Big Data Framework for Online Trajectory Prediction}, author = {P Petrou and et al.}, year = {2019}, date = {2019-01-01}, booktitle = {SSTD 2019}, keywords = {}, pubstate = {published}, tppubtype = {conference} } |
Spatharis, C; et al., Collaborative multiagent reinforcement learning schemes for air traffic management Conference IISA 2019, 2019. @conference{349, title = {Collaborative multiagent reinforcement learning schemes for air traffic management}, author = {C Spatharis and et al.}, year = {2019}, date = {2019-01-01}, booktitle = {IISA 2019}, keywords = {}, pubstate = {published}, tppubtype = {conference} } |
Vouros, G; Vlachou, A; Doulkeridis, C; Glenis, A; Santipantakis, G Efficient Spatio-temporal RDF Query Processing in Large Dynamic Knowledge Bases Conference SAC 2019, 2019. @conference{346, title = {Efficient Spatio-temporal RDF Query Processing in Large Dynamic Knowledge Bases}, author = {G Vouros and A Vlachou and C Doulkeridis and A Glenis and G Santipantakis}, year = {2019}, date = {2019-01-01}, booktitle = {SAC 2019}, keywords = {}, pubstate = {published}, tppubtype = {conference} } |
Doulkeridis, Christos; Qu, Qiang; Vouros, George A; ~a, Jo Guest Editorial: Special issue on mobility analytics for spatio-temporal and social data. Journal Article GEOINFORMATICA, 23 , 2019. @article{353, title = {Guest Editorial: Special issue on mobility analytics for spatio-temporal and social data.}, author = {Christos Doulkeridis and Qiang Qu and George A Vouros and Jo ~a}, url = {https://link.springer.com/article/10.1007%2Fs10707-019-00374-x}, year = {2019}, date = {2019-01-01}, journal = {GEOINFORMATICA}, volume = {23}, chapter = {235}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
Koutroumanis, N; et al., Integration of Mobility Data with Weather Information. Conference EDBT/ICDT Workshops 2019, 2019. @conference{348, title = {Integration of Mobility Data with Weather Information.}, author = {N Koutroumanis and et al.}, year = {2019}, date = {2019-01-01}, booktitle = {EDBT/ICDT Workshops 2019}, keywords = {}, pubstate = {published}, tppubtype = {conference} } |
Santipantakis, G; et al., stLD: towards a spatio-temporal link discovery framework. Conference SBD@SIGMOD 2019, 2019. @conference{347, title = {stLD: towards a spatio-temporal link discovery framework.}, author = {G Santipantakis and et al.}, year = {2019}, date = {2019-01-01}, booktitle = {SBD@SIGMOD 2019}, keywords = {}, pubstate = {published}, tppubtype = {conference} } |
2018 |
Santipantakis, G; Doulkeridis, C; Vouros, G; Vlachou, A MaskLink: Efficient Link Discovery for Spatial Relations via Masking Areas Journal Article arxiv, 2018. @article{336, title = {MaskLink: Efficient Link Discovery for Spatial Relations via Masking Areas}, author = {G Santipantakis and C Doulkeridis and G Vouros and A Vlachou}, url = {http://arxiv.org/abs/1803.01135}, year = {2018}, date = {2018-03-01}, journal = {arxiv}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
Vouros, George A; et al., Big Data Analytics for Time Critical Mobility Forecasting: Recent Progress and Research Challenges Proceeding 2018. @proceedings{329, title = {Big Data Analytics for Time Critical Mobility Forecasting: Recent Progress and Research Challenges}, author = {George A Vouros and et al.}, year = {2018}, date = {2018-01-01}, journal = {EDBT 2018}, keywords = {}, pubstate = {published}, tppubtype = {proceedings} } |
Nikitopoulos, P; Vlachou, A; Doulkeridis, C; Vouros, G A DiStRDF: Distributed Spatio-temporal RDF Queries on Spark Proceeding 2018. @proceedings{335, title = {DiStRDF: Distributed Spatio-temporal RDF Queries on Spark}, author = {P Nikitopoulos and A Vlachou and C Doulkeridis and G A Vouros}, year = {2018}, date = {2018-01-01}, journal = {Proceedings of the Workshops of the EDBT/ICDT 2018 Joint Conference}, keywords = {}, pubstate = {published}, tppubtype = {proceedings} } |
Santipantakis, Georgios M; et al., FAIMUSS: Flexible Data Transformation to RDF from Multiple Streaming Sources Proceeding 2018. @proceedings{331, title = {FAIMUSS: Flexible Data Transformation to RDF from Multiple Streaming Sources}, author = {Georgios M Santipantakis and et al.}, year = {2018}, date = {2018-01-01}, journal = {EDBT 2018}, keywords = {}, pubstate = {published}, tppubtype = {proceedings} } |
Kofinas, Panagiotis; Dounis, Anastasios; Vouros, George A Fuzzy Q-Learning for Multi-Agent Decentralized Energy Management in Microgrids Journal Article Applied Energy, (accepted) , 2018. @article{332, title = {Fuzzy Q-Learning for Multi-Agent Decentralized Energy Management in Microgrids}, author = {Panagiotis Kofinas and Anastasios Dounis and George A Vouros}, year = {2018}, date = {2018-01-01}, journal = {Applied Energy}, volume = {(accepted)}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
Vouros, George A; et al., Increasing Maritime Situation Awareness via Trajectory Detection, Enrichment and Recognition of Events Proceeding 2018. @proceedings{330, title = {Increasing Maritime Situation Awareness via Trajectory Detection, Enrichment and Recognition of Events}, author = {George A Vouros and et al.}, year = {2018}, date = {2018-01-01}, journal = {W2GIS 2018}, keywords = {}, pubstate = {published}, tppubtype = {proceedings} } |
Doulkeridis, Christos; Vouros, George A; Qu, Qiang; Wang, Shuhui Springer, 10731 , 2018, ISBN: 978-3-319-73520-7. @proceedings{333, title = {Mobility Analytics for Spatio-Temporal and Social Data – First International Workshop, MATES 2017, Munich, Germany, September 1, 2017, Revised Selected Papers.}, author = {Christos Doulkeridis and George A Vouros and Qiang Qu and Shuhui Wang}, url = {http://dblp.uni-trier.de/db/conf/vldb/mates2017.html}, isbn = {978-3-319-73520-7}, year = {2018}, date = {2018-01-01}, volume = {10731}, publisher = {Springer}, keywords = {}, pubstate = {published}, tppubtype = {proceedings} } |
Calvo, Esther; Cordero, Jose Manuel; Chalkiadakis, Georgios; Spatharis, Christos; Kravaris, Theocharis; Vouros, George; Blekas, Konstantinos Multiagent Reinforcement Learning Methods to Resolve Demand Capacity Balance Problems Proceeding 2018. @proceedings{334, title = {Multiagent Reinforcement Learning Methods to Resolve Demand Capacity Balance Problems}, author = {Esther Calvo and Jose Manuel Cordero and Georgios Chalkiadakis and Christos Spatharis and Theocharis Kravaris and George Vouros and Konstantinos Blekas}, year = {2018}, date = {2018-01-01}, journal = {Hellenic Artificial Intelligence Conference (SETN)}, keywords = {}, pubstate = {published}, tppubtype = {proceedings} } |
Santipantakis, Georgios; Doulkeridis, Christos; Kotis, Konstantinos; Vouros, George A RDF-Gen: Generating RDF from streaming and archival data Proceeding Novi Sad, Serbia, 2018. @proceedings{338, title = {RDF-Gen: Generating RDF from streaming and archival data}, author = {Georgios Santipantakis and Christos Doulkeridis and Konstantinos Kotis and George A Vouros}, year = {2018}, date = {2018-01-01}, journal = {WIMS 2018}, address = {Novi Sad, Serbia}, keywords = {}, pubstate = {published}, tppubtype = {proceedings} } |
Vouros, George A; Kontopoulos, Ioannis; Santipantakis, George M; Vlachou, Akrivi; Doulkeridis, Christos; Artikis, Alexander A Stream Reasoning System for Maritime Monitoring Proceeding 2018. @proceedings{341, title = {A Stream Reasoning System for Maritime Monitoring}, author = {George A Vouros and Ioannis Kontopoulos and George M Santipantakis and Akrivi Vlachou and Christos Doulkeridis and Alexander Artikis}, year = {2018}, date = {2018-01-01}, journal = {TIME 2018}, keywords = {}, pubstate = {published}, tppubtype = {proceedings} } |
Karampelas, Andreas; Vouros, George A Time and Space Efficient Large-Scale Link Discovery using String Similarities Conference WIMS 2018, Novi Sad, Serbia, 2018. @conference{339, title = {Time and Space Efficient Large-Scale Link Discovery using String Similarities}, author = {Andreas Karampelas and George A Vouros}, year = {2018}, date = {2018-01-01}, booktitle = {WIMS 2018}, address = {Novi Sad, Serbia}, keywords = {}, pubstate = {published}, tppubtype = {conference} } |
Spatharis, Christos; Kravaris, Theocharis; Blekas, Konstantinos; Vouros, George A Multiagent Reinforcement Learning Methods for Resolving Demand – Capacity Imbalances Conference DASC 2018, London, UK, 2018. @conference{337, title = {Multiagent Reinforcement Learning Methods for Resolving Demand – Capacity Imbalances}, author = {Christos Spatharis and Theocharis Kravaris and Konstantinos Blekas and George A Vouros}, year = {2018}, date = {2018-00-01}, booktitle = {DASC 2018}, address = {London, UK}, keywords = {}, pubstate = {published}, tppubtype = {conference} } |
Koubarakis, Manolis; Vouros, George; et. al., AI in Greece: The Case of Research on Linked Geospatial Data Journal Article AI Magazine, 39 , 2018. @article{340, title = {AI in Greece: The Case of Research on Linked Geospatial Data}, author = {Manolis Koubarakis and George Vouros and et. al.}, year = {2018}, date = {2018-00-01}, journal = {AI Magazine}, volume = {39}, chapter = {91}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
Cordero, J M; Herranz, R; Scarlatti, D; Marcos, R; Vouros, G; Andrienko, Gennady; Andrienko, Natalia; Fuchs, Georg; Rueping, S Visual Analytics of Flight Trajectories for Uncovering Decision Making Strategies Conference SESAR Innovation Days (SID) 2018, 2018. @conference{343, title = {Visual Analytics of Flight Trajectories for Uncovering Decision Making Strategies}, author = {J M Cordero and R Herranz and D Scarlatti and R Marcos and G Vouros and Gennady Andrienko and Natalia Andrienko and Georg Fuchs and S Rueping}, year = {2018}, date = {2018-00-01}, booktitle = {SESAR Innovation Days (SID) 2018}, keywords = {}, pubstate = {published}, tppubtype = {conference} } |
Search with keywords
| 1. | George Papadopoulos Alevizos Bastas, George Vouros Ian Crook Natalia Andrienko Gennady Andrienko Jose Manuel Cordero A: Deep reinforcement learning in service of air traffic controllers to resolve tactical conflicts. In: Expert Systems with Applications, 236 , 2024, ISBN: 0957-4174. (Type: Journal Article | Abstract | Links | BibTeX) @article{Papadopoulos2024, title = {Deep reinforcement learning in service of air traffic controllers to resolve tactical conflicts}, author = {George Papadopoulos, Alevizos Bastas, George A. Vouros, Ian Crook, Natalia Andrienko, Gennady Andrienko, Jose Manuel Cordero }, doi = {https://doi.org/10.1016/j.eswa.2023.121234}, isbn = { 0957-4174}, year = {2024}, date = {2024-02-01}, journal = {Expert Systems with Applications}, volume = {236}, abstract = {Dense and complex air traffic requires higher levels of automation than those exhibited by tactical conflict detection and resolution (CD&R) tools that air traffic controllers (ATCOs) use today: AI tools can act on their own initiative, increasing the capacity of ATCOs to control higher volumes of traffic. However, given that the air traffic control (ATC) domain is safety critical, requires AI systems to which ATCOs are comfortable to relinquishing control, guaranteeing operational integrity and automation adoption. Two major factors towards this goal are quality of solutions and operational transparency. ResoLver, the system that this article presents, addresses these challenges using an enhanced graph convolutional reinforcement learning method operating in a multiagent setting where each agent – representing a flight – performs a CD&R task, jointly with other agents. We show that ResoLver can provide high-quality solutions with respect to stakeholders interests (air traffic controllers and airspace users), addressing also operational transparency issues, which have been validated by ATCOs in simulated real-world settings.}, keywords = {}, pubstate = {published}, tppubtype = {article} } Dense and complex air traffic requires higher levels of automation than those exhibited by tactical conflict detection and resolution (CD&R) tools that air traffic controllers (ATCOs) use today: AI tools can act on their own initiative, increasing the capacity of ATCOs to control higher volumes of traffic. However, given that the air traffic control (ATC) domain is safety critical, requires AI systems to which ATCOs are comfortable to relinquishing control, guaranteeing operational integrity and automation adoption. Two major factors towards this goal are quality of solutions and operational transparency. ResoLver, the system that this article presents, addresses these challenges using an enhanced graph convolutional reinforcement learning method operating in a multiagent setting where each agent – representing a flight – performs a CD&R task, jointly with other agents. We show that ResoLver can provide high-quality solutions with respect to stakeholders interests (air traffic controllers and airspace users), addressing also operational transparency issues, which have been validated by ATCOs in simulated real-world settings. |
| 2. | Antonio Gracia-Berna Jose Manuel Cordero, Natividad Valle Ruben Rodriguez Gennady Andrienko Natalia Andrienko George Vouros Ian Crook Sandrine Molton A: Framework for transparent, explainable and trustworthy automation of ATM. DASC 2023, Forthcoming. (Type: Conference | Abstract | BibTeX) @conference{Gracia-Berna´2023, title = {Framework for transparent, explainable and trustworthy automation of ATM}, author = {Antonio Gracia-Berna, Jose Manuel Cordero, Natividad Valle, Ruben Rodriguez, Gennady Andrienko, Natalia Andrienko, George A. Vouros, Ian Crook, Sandrine Molton}, year = {2023}, date = {2023-06-20}, booktitle = {DASC 2023}, abstract = {Scientific studies before the COVID-19 pandemic indicated that Air Traffic Management (ATM) was close to saturation. The integration of Artificial Intelligence (AI) into ATM has been identified as crucial to achieve higher levels of automation, and the need for trustworthy and explainable automation systems in safety-critical domains is essential. Boeing Aerospace Spain (BAS) has conducted pioneering research on achieving high levels of automation while ensuring transparency and explainability. BAS, along with other major ATM players, has developed a framework for implementing transparent and explainable automation using Explainable Artificial Intelligence (XAI) for Air Traffic Flow and Capacity Management (ATFCM), and Conflict Detection and Resolution (CDR) scenarios. The framework provides a set of principles for the transparent application of XAI technology in ATM to ensure that different types of audiences can trust the AI system’s decisions. The principles have been developed based on feedback from experts in ATM, human factors, and AI. This framework can be considered the first attempt to pave the way for AI techniques to achieve higher levels of automation in accordance with the European ATM Master Plan.}, keywords = {}, pubstate = {forthcoming}, tppubtype = {conference} } Scientific studies before the COVID-19 pandemic indicated that Air Traffic Management (ATM) was close to saturation. The integration of Artificial Intelligence (AI) into ATM has been identified as crucial to achieve higher levels of automation, and the need for trustworthy and explainable automation systems in safety-critical domains is essential. Boeing Aerospace Spain (BAS) has conducted pioneering research on achieving high levels of automation while ensuring transparency and explainability. BAS, along with other major ATM players, has developed a framework for implementing transparent and explainable automation using Explainable Artificial Intelligence (XAI) for Air Traffic Flow and Capacity Management (ATFCM), and Conflict Detection and Resolution (CDR) scenarios. The framework provides a set of principles for the transparent application of XAI technology in ATM to ensure that different types of audiences can trust the AI system’s decisions. The principles have been developed based on feedback from experts in ATM, human factors, and AI. This framework can be considered the first attempt to pave the way for AI techniques to achieve higher levels of automation in accordance with the European ATM Master Plan. |
| 3. | Georgios Santipantakis Christos Doulkeridis, George Vouros A: An Ontology for Representing and Querying Semantic Trajectories in the Maritime Domain. ADBIS 2023, Forthcoming. (Type: Conference | BibTeX) @conference{Santipantakis2023, title = {An Ontology for Representing and Querying Semantic Trajectories in the Maritime Domain}, author = {Georgios Santipantakis, Christos Doulkeridis, George A. Vouros}, year = {2023}, date = {2023-06-20}, booktitle = {ADBIS 2023}, keywords = {}, pubstate = {forthcoming}, tppubtype = {conference} } |
| 4. | Alevizos Bastas, George Vouros A: Data-Driven Modeling of Air Traffic Controllers’ Policy to Resolve Conflicts. In: Aerospace, 10 (6), 2023, ISSN: 2226-4310. (Type: Journal Article | Abstract | Links | BibTeX) @article{Bastas2023, title = {Data-Driven Modeling of Air Traffic Controllers’ Policy to Resolve Conflicts}, author = {Alevizos Bastas, George A. Vouros}, doi = {https://doi.org/10.3390/aerospace10060557}, issn = {2226-4310}, year = {2023}, date = {2023-06-13}, journal = {Aerospace}, volume = {10}, number = {6}, abstract = {With the aim to enhance automation in conflict detection and resolution (CD&R) tasks in the air traffic management (ATM) domain, this article studies the use of artificial intelligence and machine learning (AI/ML) methods to learn air traffic controllers’ (ATCOs) policy in resolving conflicts among aircraft assessed to violate separation minimum constraints during the en route phase of flights, in the tactical phase of operations. The objective is to model ℎ𝑜𝑤 conflicts are being resolved by ATCOs. Towards this goal, the article formulates the ATCO policy learning problem for conflict resolution, addresses the challenging issue of an inherent lack of information in real-world data, and presents AI/ML methods that learn models of ATCOs’ behavior. The methods are evaluated using real-world datasets. The results show that AI/ML methods can achieve good accuracy on predicting ATCOs’ actions given specific conflicts, revealing the preferences of ATCOs for resolution actions in specific circumstances. However, the high accuracy of predictions is hindered by real-world data-inherent limitations.}, keywords = {}, pubstate = {published}, tppubtype = {article} } With the aim to enhance automation in conflict detection and resolution (CD&R) tasks in the air traffic management (ATM) domain, this article studies the use of artificial intelligence and machine learning (AI/ML) methods to learn air traffic controllers’ (ATCOs) policy in resolving conflicts among aircraft assessed to violate separation minimum constraints during the en route phase of flights, in the tactical phase of operations. The objective is to model ℎ𝑜𝑤 conflicts are being resolved by ATCOs. Towards this goal, the article formulates the ATCO policy learning problem for conflict resolution, addresses the challenging issue of an inherent lack of information in real-world data, and presents AI/ML methods that learn models of ATCOs’ behavior. The methods are evaluated using real-world datasets. The results show that AI/ML methods can achieve good accuracy on predicting ATCOs’ actions given specific conflicts, revealing the preferences of ATCOs for resolution actions in specific circumstances. However, the high accuracy of predictions is hindered by real-world data-inherent limitations. |
| 5. | Andreas Kontogiannis, George Vouros : Inherently Interpretable Deep Reinforcement Learning through Online Mimicking. In: EXTRAAMAS @AAMAS 2023, 2023. (Type: Inproceedings | Abstract | BibTeX) @inproceedings{Kontogiannis2023, title = {Inherently Interpretable Deep Reinforcement Learning through Online Mimicking}, author = {Andreas Kontogiannis, George Vouros}, year = {2023}, date = {2023-05-29}, booktitle = {EXTRAAMAS @AAMAS 2023}, abstract = {Although deep reinforcement learning (DRL) methods have been successfully applied in challenging tasks, their application in real-world operational settings – where transparency and accountability play important roles in automation – is challenged by methods’ limited ability to provide explanations. Among the paradigms for explainability in DRL is the interpretable box design paradigm, where interpretable models substitute inner closed constituent models of the DRL method, thus making the DRL method “inherently” interpretable. In this paper we propose a generic paradigm where interpretable policy models are trained following an online mimicking paradigm. We exemplify this paradigm through XDQN, an explainable variation of DQN that uses an interpretable policy model trained online with the deep policy model. XDQN is challenged in a complex, real-world operational multi-agent problem pertaining to the demand-capacity balancing problem of air traffic management (ATM), where human operators need to master complexity and understand the factors driving decision making. XDQN is shown to achieve performance similar to that of its DQN counterpart, while its abilities to provide global models’ interpretations and interpretations of local decisions are demonstrated.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Although deep reinforcement learning (DRL) methods have been successfully applied in challenging tasks, their application in real-world operational settings – where transparency and accountability play important roles in automation – is challenged by methods’ limited ability to provide explanations. Among the paradigms for explainability in DRL is the interpretable box design paradigm, where interpretable models substitute inner closed constituent models of the DRL method, thus making the DRL method “inherently" interpretable. In this paper we propose a generic paradigm where interpretable policy models are trained following an online mimicking paradigm. We exemplify this paradigm through XDQN, an explainable variation of DQN that uses an interpretable policy model trained online with the deep policy model. XDQN is challenged in a complex, real-world operational multi-agent problem pertaining to the demand-capacity balancing problem of air traffic management (ATM), where human operators need to master complexity and understand the factors driving decision making. XDQN is shown to achieve performance similar to that of its DQN counterpart, while its abilities to provide global models’ interpretations and interpretations of local decisions are demonstrated. |
| 6. | Theocharis Kravaris Konstantinos Lentzos, Georgios Santipantakis George Vouros Gennady Andrienko Natalia Andrienko Ian Crook Jose Manuel Cordero Garcia Enrique Iglesias Martinez A L V: Explaining deep reinforcement learning decisions in complex multiagent settings: towards enabling automation in air traffic flow management. In: Applied Intelligence, 53 , pp. 4063-4098, 2023. (Type: Journal Article | Abstract | Links | BibTeX) @article{Kravaris2023, title = {Explaining deep reinforcement learning decisions in complex multiagent settings: towards enabling automation in air traffic flow management}, author = {Theocharis Kravaris, Konstantinos Lentzos, Georgios Santipantakis, George A. Vouros, Gennady L. Andrienko, Natalia V. Andrienko, Ian Crook, Jose Manuel Cordero Garcia, Enrique Iglesias Martinez}, doi = {10.1007/s10489-022-03605-1}, year = {2023}, date = {2023-01-20}, journal = {Applied Intelligence}, volume = {53}, pages = {4063-4098}, abstract = {With the objective to enhance human performance and maximize engagement during the performance of tasks, we aim to advance automation for decision making in complex and large-scale multi-agent settings. Towards these goals, this paper presents a deep multi agent reinforcement learning method for resolving demand – capacity imbalances in real-world Air Traffic Management settings with thousands of agents. Agents comprising the system are able to jointly decide on the measures to be applied to resolve imbalances, while they provide explanations on their decisions: This information is rendered and explored via appropriate visual analytics tools. The paper presents how major challenges of scalability and complexity are addressed, and provides results from evaluation tests that show the abilities of models to provide high-quality solutions and high-fidelity explanations.}, keywords = {}, pubstate = {published}, tppubtype = {article} } With the objective to enhance human performance and maximize engagement during the performance of tasks, we aim to advance automation for decision making in complex and large-scale multi-agent settings. Towards these goals, this paper presents a deep multi agent reinforcement learning method for resolving demand – capacity imbalances in real-world Air Traffic Management settings with thousands of agents. Agents comprising the system are able to jointly decide on the measures to be applied to resolve imbalances, while they provide explanations on their decisions: This information is rendered and explored via appropriate visual analytics tools. The paper presents how major challenges of scalability and complexity are addressed, and provides results from evaluation tests that show the abilities of models to provide high-quality solutions and high-fidelity explanations. |
| 7. | Christos Spatharis Alevizos Bastas, Theocharis Kravaris Konstantinos Blekas George Vouros Jose Manuel Cordero Garcia A: Hierarchical multiagent reinforcement learning schemes for air traffic management. In: Neural Computing and Applications, 35 , pp. 147-159, 2023. (Type: Journal Article | Abstract | Links | BibTeX) @article{Spatharis2023, title = {Hierarchical multiagent reinforcement learning schemes for air traffic management}, author = {Christos Spatharis, Alevizos Bastas, Theocharis Kravaris, Konstantinos Blekas, George A. Vouros, Jose Manuel Cordero Garcia}, doi = {10.1007/s00521-021-05748-7}, year = {2023}, date = {2023-01-01}, journal = {Neural Computing and Applications}, volume = {35}, pages = {147-159}, abstract = {In this work we investigate the use of hierarchical multiagent reinforcement learning methods for the computation of policies to resolve congestion problems in the air traffic management domain. To address cases where the demand of airspace use exceeds capacity, we consider agents representing flights, who need to decide on ground delays at the pre-tactical stage of operations, towards executing their trajectories while adhering to airspace capacity constraints. Hierarchical reinforcement learning manages to handle real-world problems with high complexity, by partitioning the task into hierarchies of states and/or actions. This provides an efficient way of exploring the state–action space and constructing an advantageous decision-making mechanism. We first establish a general framework of hierarchical multiagent reinforcement learning, and then, we further formulate four alternative schemes of abstractions, on states, actions, or both. To quantitatively assess the quality of solutions of the proposed approaches and show the potential of the hierarchical methods in resolving the demand–capacity balance problem, we provide experimental results on real-world evaluation cases, where we measure the average delay per flight and the number of flights with delays.}, keywords = {}, pubstate = {published}, tppubtype = {article} } In this work we investigate the use of hierarchical multiagent reinforcement learning methods for the computation of policies to resolve congestion problems in the air traffic management domain. To address cases where the demand of airspace use exceeds capacity, we consider agents representing flights, who need to decide on ground delays at the pre-tactical stage of operations, towards executing their trajectories while adhering to airspace capacity constraints. Hierarchical reinforcement learning manages to handle real-world problems with high complexity, by partitioning the task into hierarchies of states and/or actions. This provides an efficient way of exploring the state–action space and constructing an advantageous decision-making mechanism. We first establish a general framework of hierarchical multiagent reinforcement learning, and then, we further formulate four alternative schemes of abstractions, on states, actions, or both. To quantitatively assess the quality of solutions of the proposed approaches and show the potential of the hierarchical methods in resolving the demand–capacity balance problem, we provide experimental results on real-world evaluation cases, where we measure the average delay per flight and the number of flights with delays. |
| 8. | Vouros, George A: Explainable Deep Reinforcement Learning: State of the Art and Challenges. In: ACM Computing Surveys, 55 (5), pp. 1-39, 2023. (Type: Journal Article | Abstract | Links | BibTeX) @article{Vouros2023, title = {Explainable Deep Reinforcement Learning: State of the Art and Challenges}, author = {George A. Vouros}, doi = {10.1145/3527448}, year = {2023}, date = {2023-01-01}, journal = {ACM Computing Surveys}, volume = {55}, number = {5}, pages = {1-39}, abstract = {Interpretability, explainability, and transparency are key issues to introducing artificial intelligence methods in many critical domains. This is important due to ethical concerns and trust issues strongly connected to reliability, robustness, auditability, and fairness, and has important consequences toward keeping the human in the loop in high levels of automation, especially in critical cases for decision making, where both (human and the machine) play important roles. Although the research community has given much attention to explainability of closed (or black) prediction boxes, there are tremendous needs for explainability of closed-box methods that support agents to act autonomously in the real world. Reinforcement learning methods, and especially their deep versions, are such closed-box methods. In this article, we aim to provide a review of state-of-the-art methods for explainable deep reinforcement learning methods, taking also into account the needs of human operators—that is, of those who make the actual and critical decisions in solving real-world problems. We provide a formal specification of the deep reinforcement learning explainability problems, and we identify the necessary components of a general explainable reinforcement learning framework. Based on these, we provide a comprehensive review of state-of-the-art methods, categorizing them into classes according to the paradigm they follow, the interpretable models they use, and the surface representation of explanations provided. The article concludes by identifying open questions and important challenges.}, keywords = {}, pubstate = {published}, tppubtype = {article} } Interpretability, explainability, and transparency are key issues to introducing artificial intelligence methods in many critical domains. This is important due to ethical concerns and trust issues strongly connected to reliability, robustness, auditability, and fairness, and has important consequences toward keeping the human in the loop in high levels of automation, especially in critical cases for decision making, where both (human and the machine) play important roles. Although the research community has given much attention to explainability of closed (or black) prediction boxes, there are tremendous needs for explainability of closed-box methods that support agents to act autonomously in the real world. Reinforcement learning methods, and especially their deep versions, are such closed-box methods. In this article, we aim to provide a review of state-of-the-art methods for explainable deep reinforcement learning methods, taking also into account the needs of human operators—that is, of those who make the actual and critical decisions in solving real-world problems. We provide a formal specification of the deep reinforcement learning explainability problems, and we identify the necessary components of a general explainable reinforcement learning framework. Based on these, we provide a comprehensive review of state-of-the-art methods, categorizing them into classes according to the paradigm they follow, the interpretable models they use, and the surface representation of explanations provided. The article concludes by identifying open questions and important challenges. |
| 9. | Natividad Valle M Florencia Lema, José Manuel Cordero Enrique Iglesias Rubén Rodríguez George Vouros Theocharis Kravaris George Papadopoulos Alevizos Bastas Georgios Santipantakis A: Explainability & Transparency in higher levels of automation in the ATM domain. In: SESAR Innovation Days 2022, 2022. (Type: Inproceedings | Abstract | Links | BibTeX) @inproceedings{Valle2022, title = {Explainability & Transparency in higher levels of automation in the ATM domain}, author = {Natividad Valle, M Florencia Lema, José Manuel Cordero, Enrique Iglesias , Rubén Rodríguez, George A. Vouros, Theocharis Kravaris, George Papadopoulos, Alevizos Bastas, Georgios Santipantakis}, url = {https://www.sesarju.eu/sesarinnovationdays}, year = {2022}, date = {2022-12-06}, booktitle = {SESAR Innovation Days 2022}, journal = {IEEE Comput Graph Appl.}, abstract = {This paper presents findings, lessons learnt and guidelines for the use of explainable and transparent Artificial Intelligence (AI)/Machine Learning (ML) in ATM. The paper focuses on the results obtained from validating two AI/ML prototypes for Conflict Detection & Resolution (CD&R) and Air Traffic Flow Capacity Management (ATFCM) problems. These two prototypes are representative of the type of advanced automated systems that can support respectively the tactical and the pre-tactical operational phases The aim is, shifting the paradigm of human-AI teaming, providing full explainability and operational transparency. The major question is: when and how explanations should be provided for systems to be acceptable and trustworthy by operators?}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } This paper presents findings, lessons learnt and guidelines for the use of explainable and transparent Artificial Intelligence (AI)/Machine Learning (ML) in ATM. The paper focuses on the results obtained from validating two AI/ML prototypes for Conflict Detection & Resolution (CD&R) and Air Traffic Flow Capacity Management (ATFCM) problems. These two prototypes are representative of the type of advanced automated systems that can support respectively the tactical and the pre-tactical operational phases The aim is, shifting the paradigm of human-AI teaming, providing full explainability and operational transparency. The major question is: when and how explanations should be provided for systems to be acceptable and trustworthy by operators? |
| 10. | George A. Vouros Theodore Tranos, Konstantinos Blekas Georgios Santipantakis Marc Melgosa Xavier Prats : Data-driven estimation of flights’ hidden parameters . In: SESAR Innovation Days 2022, 2022. (Type: Inproceedings | Abstract | Links | BibTeX) @inproceedings{Vouros2022c, title = {Data-driven estimation of flights’ hidden parameters }, author = {George A. Vouros, Theodore Tranos, Konstantinos Blekas, Georgios Santipantakis, Marc Melgosa, Xavier Prats }, url = {https://www.sesarju.eu/sesarinnovationdays}, year = {2022}, date = {2022-12-06}, booktitle = {SESAR Innovation Days 2022}, abstract = {This paper presents a data-driven methodology for the estimation of flights’ hidden parameters, combining mechanistic and AI/ML models. In the context of this methodology the paper studies several AI/ML methods and reports on evaluation results for estimating hidden parameters, in terms of mean absolute error. In addition to the estimation of hidden parameters themselves, this paper examines how these estimations affect the prediction of KPIs regarding the efficiency of flights using a mechanistic model. Results show the accuracy of the proposed methods and the benefits of the proposed methodology. Indeed, the results show significant advances of data-driven methods to estimate hidden parameters towards predicting KPIs.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } This paper presents a data-driven methodology for the estimation of flights’ hidden parameters, combining mechanistic and AI/ML models. In the context of this methodology the paper studies several AI/ML methods and reports on evaluation results for estimating hidden parameters, in terms of mean absolute error. In addition to the estimation of hidden parameters themselves, this paper examines how these estimations affect the prediction of KPIs regarding the efficiency of flights using a mechanistic model. Results show the accuracy of the proposed methods and the benefits of the proposed methodology. Indeed, the results show significant advances of data-driven methods to estimate hidden parameters towards predicting KPIs. |
| 11. | Alevizos Bastas, George Vouros A: Data-driven prediction of Air Traffic Controllers reactions to resolving conflicts. In: Information Sciences, 613 , pp. 763-785, 2022. (Type: Journal Article | Abstract | Links | BibTeX) @article{Bastas2022, title = {Data-driven prediction of Air Traffic Controllers reactions to resolving conflicts}, author = {Alevizos Bastas, George A. Vouros}, doi = {10.1016/j.ins.2022.09.015}, year = {2022}, date = {2022-10-01}, journal = {Information Sciences}, volume = {613}, pages = {763-785}, abstract = {With the aim to enhance automation in conflict detection and resolution (CD&R) tasks in the Air Traffic Management domain, this article proposes deep learning (DL) techniques that model Air Traffic Controllers’ reactions in resolving conflicts violating aircraft trajectories separation minimum constraints: This implies learning when the Air Traffic Controller reacts towards resolving a conflict, and how he/she reacts. Timely reactions, to which this article aims, focus on when do reactions happen, aiming to predict the trajectory points, as the aircraft state evolves, that the Air Traffic Controller (ATCO) issues a conflict resolution action. Towards this goal, the article formulates the Air Traffic Controllers’ reaction prediction problem for CD&R, presents DL methods that can model Air Traffic Controllers’ timely reactions, and evaluates these methods in real-world data sets, showing their efficacy in solving the problem with very high accuracy.}, keywords = {}, pubstate = {published}, tppubtype = {article} } With the aim to enhance automation in conflict detection and resolution (CD&R) tasks in the Air Traffic Management domain, this article proposes deep learning (DL) techniques that model Air Traffic Controllers’ reactions in resolving conflicts violating aircraft trajectories separation minimum constraints: This implies learning when the Air Traffic Controller reacts towards resolving a conflict, and how he/she reacts. Timely reactions, to which this article aims, focus on when do reactions happen, aiming to predict the trajectory points, as the aircraft state evolves, that the Air Traffic Controller (ATCO) issues a conflict resolution action. Towards this goal, the article formulates the Air Traffic Controllers’ reaction prediction problem for CD&R, presents DL methods that can model Air Traffic Controllers’ timely reactions, and evaluates these methods in real-world data sets, showing their efficacy in solving the problem with very high accuracy. |
| 12. | Vouros, George A: Tutorial on Explainable Deep Reinforcement Learning: One framework, three paradigms and many challenges. In: SETN 2022, ACM, 2022. (Type: Inproceedings | Abstract | Links | BibTeX) @inproceedings{Vouros2022, title = {Tutorial on Explainable Deep Reinforcement Learning: One framework, three paradigms and many challenges}, author = {George A. Vouros}, doi = {10.1145/3549737.3549808}, year = {2022}, date = {2022-09-10}, booktitle = {SETN 2022}, publisher = {ACM}, abstract = {Interpretability, explainability and transparency are key issues to introducing Artificial Intelligence closed—box methods in many critical domains: This is important due to ethical concerns and trust issues strongly connected to reliability, robustness, auditability and fairness, and has important consequences towards keeping the human in the loop in high levels of automation, especially in critical cases for decision making. Reinforcement learning methods, and especially their deep versions, are closed-box methods that support agents to act autonomously in the real world. This tutorial will provide a formal specification of the deep reinforcement learning explainability problems, and will present the necessary components of a general explainable reinforcement learning framework. Based on this framework will provide distinct explainability paradigms towards solving explainability problems, with examples from state-of-the-art methods and real-world cases. The tutorial will conclude identifying open questions and important challenges. The tutorial is based on the survey paper on “Explainable Deep Reinforcement Learning” State of the Art and Challenges”}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Interpretability, explainability and transparency are key issues to introducing Artificial Intelligence closed—box methods in many critical domains: This is important due to ethical concerns and trust issues strongly connected to reliability, robustness, auditability and fairness, and has important consequences towards keeping the human in the loop in high levels of automation, especially in critical cases for decision making. Reinforcement learning methods, and especially their deep versions, are closed-box methods that support agents to act autonomously in the real world. This tutorial will provide a formal specification of the deep reinforcement learning explainability problems, and will present the necessary components of a general explainable reinforcement learning framework. Based on this framework will provide distinct explainability paradigms towards solving explainability problems, with examples from state-of-the-art methods and real-world cases. The tutorial will conclude identifying open questions and important challenges. The tutorial is based on the survey paper on “Explainable Deep Reinforcement Learning” State of the Art and Challenges" |
| 13. | Apostolos Glenis, George Vouros A: SCALE-BOSS: A framework for scalable time-series classification using symbolic representations. In: SETN 2022, ACM, 2022. (Type: Inproceedings | Abstract | Links | BibTeX) @inproceedings{Glenis2022, title = {SCALE-BOSS: A framework for scalable time-series classification using symbolic representations}, author = {Apostolos Glenis, George A. Vouros}, doi = {10.1145/3549737.3549761}, year = {2022}, date = {2022-09-10}, booktitle = {SETN 2022}, publisher = {ACM}, abstract = {Time-Series Classification (TSC) is an important problem in many fields across sciences. Many algorithms for TSC use symbolic representation to combat noise. In this paper we propose a framework, namely SCALE-BOSS, to build TSC algorithms that exploit time-series models based on symbolic representations. While alternative symbolic representations can be incorporated, we have opted to use the Bag-Of-SFA (BOSS) approach, and thus SFA, as a state-of-the-art symbolic time series representation. We investigate the efficiency of several instantiations of this framework based on two main variations, where the TSC model is built either by a time-series classification or by a clustering algorithm. The objective is to advance the computational efficiency of TSC classification algorithms without sacrificing their accuracy. We evaluate the instantiations of the SCALE-BOSS framework on those datasets in the UCR time-series repository that include the largest training sets. Comparisons with state of the art methods on TSC show the balance between computational efficiency and accuracy on predictions achieved.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Time-Series Classification (TSC) is an important problem in many fields across sciences. Many algorithms for TSC use symbolic representation to combat noise. In this paper we propose a framework, namely SCALE-BOSS, to build TSC algorithms that exploit time-series models based on symbolic representations. While alternative symbolic representations can be incorporated, we have opted to use the Bag-Of-SFA (BOSS) approach, and thus SFA, as a state-of-the-art symbolic time series representation. We investigate the efficiency of several instantiations of this framework based on two main variations, where the TSC model is built either by a time-series classification or by a clustering algorithm. The objective is to advance the computational efficiency of TSC classification algorithms without sacrificing their accuracy. We evaluate the instantiations of the SCALE-BOSS framework on those datasets in the UCR time-series repository that include the largest training sets. Comparisons with state of the art methods on TSC show the balance between computational efficiency and accuracy on predictions achieved. |
| 14. | Georgios M. Santipantakis Konstantinos I. Kotis, Apostolos Glenis George Vouros Christos Doulkeridis Akrivi Vlachou A: RDF-Gen: generating RDF triples from big data sources. In: Knowledge and Information Systems, 64 , pp. 2985–3015, 2022. (Type: Journal Article | Abstract | Links | BibTeX) @article{Santipantakis2022, title = {RDF-Gen: generating RDF triples from big data sources}, author = {Georgios M. Santipantakis, Konstantinos I. Kotis, Apostolos Glenis, George A. Vouros, Christos Doulkeridis, Akrivi Vlachou}, doi = {10.1007/s10115-022-01729-x}, year = {2022}, date = {2022-08-13}, journal = {Knowledge and Information Systems}, volume = {64}, pages = {2985–3015}, abstract = {Transforming disparate and heterogeneous data sources that provide large volumes of data in high velocity into a common form allows integrated and enriched views on data and thus provides further opportunities to advance the effectiveness and accuracy of data analysis and prediction tasks. This paper presents the RDF-Gen approach for transforming data provided by archival and streaming data sources, provided in various formats, into RDF triples, according to a set of ontological specifications. RDF-Gen introduces a generic mechanism which supports the transformation of data efficiently (i.e., with high throughput and low latency), even in cases where the velocity of data presents high peaks, offering facilities for discovering associations between data from different sources, and supporting transformation of modular data sets. This paper presents a parallel implementation of RDF-Gen, also presenting data transformation workflows that allow variations incorporating RDF-Gen instances, adjusting to the needs of data sources, application areas and performance requirements. RDF-Gen is experimentally evaluated against state of the art, in both archival and streaming settings: Experimental results show RDF-Gen efficiency and highlight key contributions.}, keywords = {}, pubstate = {published}, tppubtype = {article} } Transforming disparate and heterogeneous data sources that provide large volumes of data in high velocity into a common form allows integrated and enriched views on data and thus provides further opportunities to advance the effectiveness and accuracy of data analysis and prediction tasks. This paper presents the RDF-Gen approach for transforming data provided by archival and streaming data sources, provided in various formats, into RDF triples, according to a set of ontological specifications. RDF-Gen introduces a generic mechanism which supports the transformation of data efficiently (i.e., with high throughput and low latency), even in cases where the velocity of data presents high peaks, offering facilities for discovering associations between data from different sources, and supporting transformation of modular data sets. This paper presents a parallel implementation of RDF-Gen, also presenting data transformation workflows that allow variations incorporating RDF-Gen instances, adjusting to the needs of data sources, application areas and performance requirements. RDF-Gen is experimentally evaluated against state of the art, in both archival and streaming settings: Experimental results show RDF-Gen efficiency and highlight key contributions. |
| 15. | George A. Vouros George Papadopoulos, Alevizos Bastas Jose Manuel Cordero Garcia Ruben Rodrigez Rodrigez : Automating the Resolution of Flight Conflicts: Deep Reinforcement Learning in Service of Air Traffic Controllers. In: PAIS 2022, IOS PRess, 2022. (Type: Inproceedings | Abstract | Links | BibTeX) @inproceedings{Vouros2022b, title = {Automating the Resolution of Flight Conflicts: Deep Reinforcement Learning in Service of Air Traffic Controllers}, author = {George A. Vouros, George Papadopoulos, Alevizos Bastas, Jose Manuel Cordero Garcia, Ruben Rodrigez Rodrigez}, doi = {10.3233/FAIA220066}, year = {2022}, date = {2022-07-01}, booktitle = {PAIS 2022}, volume = {351}, publisher = {IOS PRess}, series = {Frontiers in Artificial Intelligence and Applications}, abstract = {Dense and complex air traffic scenarios require higher levels of automation than those exhibited by tactical conflict detection and resolution (CD&R) tools that air traffic controllers (ATCO) use today. However, the air traffic control (ATC) domain, being safety critical, requires AI systems to which operators are comfortable to relinquishing control, guaranteeing operational integrity and automation adoption. Two major factors towards this goal are quality of solutions, and transparency in decision making. This paper proposes using a graph convolutional reinforcement learning method operating in a multiagent setting where each agent (flight) performs a CD&R task, jointly with other agents. We show that this method can provide high-quality solutions with respect to stakeholders interests (air traffic controllers and airspace users), addressing operational transparency issues.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Dense and complex air traffic scenarios require higher levels of automation than those exhibited by tactical conflict detection and resolution (CD&R) tools that air traffic controllers (ATCO) use today. However, the air traffic control (ATC) domain, being safety critical, requires AI systems to which operators are comfortable to relinquishing control, guaranteeing operational integrity and automation adoption. Two major factors towards this goal are quality of solutions, and transparency in decision making. This paper proposes using a graph convolutional reinforcement learning method operating in a multiagent setting where each agent (flight) performs a CD&R task, jointly with other agents. We show that this method can provide high-quality solutions with respect to stakeholders interests (air traffic controllers and airspace users), addressing operational transparency issues. |
| 16. | Gennady L. Andrienko Natalia V. Andrienko, Jose Manuel Cordero Garcia Dirk Hecker George Vouros A: Supporting Visual Exploration of Iterative Job Scheduling. In: IEEE Comput Graph Appl . , 42 (3), pp. 74-86, 2022. (Type: Journal Article | Abstract | Links | BibTeX) @article{Andrienko2022, title = {Supporting Visual Exploration of Iterative Job Scheduling}, author = {Gennady L. Andrienko, Natalia V. Andrienko, Jose Manuel Cordero Garcia, Dirk Hecker, George A. Vouros}, doi = {10.1109/MCG.2022.3163437}, year = {2022}, date = {2022-05-06}, journal = {IEEE Comput Graph Appl . }, volume = {42}, number = {3}, pages = {74-86}, abstract = {We consider the general problem known as job shop scheduling, in which multiple jobs consist of sequential operations that need to be executed or served by appropriate machines having limited capacities. For example, train journeys (jobs) consist of moves and stops (operations) to be served by rail tracks and stations (machines). A schedule is an assignment of the job operations to machines and times where and when they will be executed. The developers of computational methods for job scheduling need tools enabling them to explore how their methods work. At a high level of generality, we define the system of pertinent exploration tasks and a combination of visualizations capable of supporting the tasks. We provide general descriptions of the purposes, contents, visual encoding, properties, and interactive facilities of the visualizations and illustrate them with images from an example implementation in air traffic management. We justify the design of the visualizations based on the tasks, principles of creating visualizations for pattern discovery, and scalability requirements. The outcomes of our research are sufficiently general to be of use in a variety of applications.}, keywords = {}, pubstate = {published}, tppubtype = {article} } We consider the general problem known as job shop scheduling, in which multiple jobs consist of sequential operations that need to be executed or served by appropriate machines having limited capacities. For example, train journeys (jobs) consist of moves and stops (operations) to be served by rail tracks and stations (machines). A schedule is an assignment of the job operations to machines and times where and when they will be executed. The developers of computational methods for job scheduling need tools enabling them to explore how their methods work. At a high level of generality, we define the system of pertinent exploration tasks and a combination of visualizations capable of supporting the tasks. We provide general descriptions of the purposes, contents, visual encoding, properties, and interactive facilities of the visualizations and illustrate them with images from an example implementation in air traffic management. We justify the design of the visualizations based on the tasks, principles of creating visualizations for pattern discovery, and scalability requirements. The outcomes of our research are sufficiently general to be of use in a variety of applications. |
| 17. | Nikolaos Koutroumanis Georgios M. Santipantakis, Apostolos Glenis Christos Doulkeridis & George Vouros A: Scalable enrichment of mobility data with weather information. In: GEOINFORMATICA, 25 , pp. 291-309, 2021. (Type: Journal Article | Abstract | Links | BibTeX) @article{Koutroumanis2021, title = {Scalable enrichment of mobility data with weather information}, author = {Nikolaos Koutroumanis, Georgios M. Santipantakis, Apostolos Glenis, Christos Doulkeridis & George A. Vouros }, doi = {10.1007/s10707-020-00423-w}, year = {2021}, date = {2021-09-17}, journal = {GEOINFORMATICA}, volume = {25}, pages = {291-309}, abstract = {More and more real-life applications for mobility analytics require the joint exploitation of positional information of moving objects together with weather data that correspond to the movement. In particular, this is evident in fleet management applications for improved routing and reduced fuel consumption, in the maritime domain for more accurate trajectory prediction, as well as in air-traffic management for predicting regulations and reducing delays. Motivated by such applications, in this paper, we present a system for the enrichment of mobility data with weather information. Our main application scenario concerns streaming positional information (such as GPS traces of vehicles) that is collected and is enriched in an online fashion with stored weather data. We present the system architecture of a centralized version that runs on a single machine and exploits caching to improve its efficiency. Also, we extend our approach to a parallel implementation on top of Apache Kafka, which can scale to hundreds of thousands of processed records when provided with more computing nodes. Furthermore, we present extensions of our system for: (a) enrichment of more complex geometries than point data, and (b) providing linked RDF data as output. Our experimental evaluation on a medium-sized cluster shows the scalability of our approach in terms of number of processed records per second.}, keywords = {}, pubstate = {published}, tppubtype = {article} } More and more real-life applications for mobility analytics require the joint exploitation of positional information of moving objects together with weather data that correspond to the movement. In particular, this is evident in fleet management applications for improved routing and reduced fuel consumption, in the maritime domain for more accurate trajectory prediction, as well as in air-traffic management for predicting regulations and reducing delays. Motivated by such applications, in this paper, we present a system for the enrichment of mobility data with weather information. Our main application scenario concerns streaming positional information (such as GPS traces of vehicles) that is collected and is enriched in an online fashion with stored weather data. We present the system architecture of a centralized version that runs on a single machine and exploits caching to improve its efficiency. Also, we extend our approach to a parallel implementation on top of Apache Kafka, which can scale to hundreds of thousands of processed records when provided with more computing nodes. Furthermore, we present extensions of our system for: (a) enrichment of more complex geometries than point data, and (b) providing linked RDF data as output. Our experimental evaluation on a medium-sized cluster shows the scalability of our approach in terms of number of processed records per second. |
| 18. | Panagiotis Nikitopoulos Akrivi Vlachou, Christos Doulkeridis George Vouros A: Parallel and scalable processing of spatio-temporal RDF queries using Spark. In: Geoinformatica, 25 , pp. 623-653, 2021. (Type: Journal Article | Abstract | Links | BibTeX) @article{Nikitopoulos2021, title = {Parallel and scalable processing of spatio-temporal RDF queries using Spark}, author = {Panagiotis Nikitopoulos, Akrivi Vlachou, Christos Doulkeridis, George A. Vouros}, doi = {10.1007/s10707-019-00371-0}, year = {2021}, date = {2021-07-03}, journal = {Geoinformatica}, volume = {25}, pages = {623-653}, abstract = {The ever-increasing size of data emanating from mobile devices and sensors, dictates the use of distributed systems for storing and querying these data. Typically, such data sources provide some spatio-temporal information, alongside other useful data. The RDF data model can be used to interlink and exchange data originating from heterogeneous sources in a uniform manner. For example, consider the case where vessels report their spatio-temporal position, on a regular basis, by using various surveillance systems. In this scenario, a user might be interested to know which vessels were moving in a specific area for a given temporal range. In this paper, we address the problem of efficiently storing and querying spatio-temporal RDF data in parallel. We specifically study the case of SPARQL queries with spatio-temporal constraints, by proposing the DiStRDF system, which is comprised of a Storage and a Processing Layer. The DiStRDF Storage Layer is responsible for efficiently storing large amount of historical spatio-temporal RDF data of moving objects. On top of it, we devise our DiStRDF Processing Layer, which parses a SPARQL query and produces corresponding logical and physical execution plans. We use Spark, a well-known distributed in-memory processing framework, as the underlying processing engine. Our experimental evaluation, on real data from both aviation and maritime domains, demonstrates the efficiency of our DiStRDF system, when using various spatio-temporal range constraints.}, keywords = {}, pubstate = {published}, tppubtype = {article} } The ever-increasing size of data emanating from mobile devices and sensors, dictates the use of distributed systems for storing and querying these data. Typically, such data sources provide some spatio-temporal information, alongside other useful data. The RDF data model can be used to interlink and exchange data originating from heterogeneous sources in a uniform manner. For example, consider the case where vessels report their spatio-temporal position, on a regular basis, by using various surveillance systems. In this scenario, a user might be interested to know which vessels were moving in a specific area for a given temporal range. In this paper, we address the problem of efficiently storing and querying spatio-temporal RDF data in parallel. We specifically study the case of SPARQL queries with spatio-temporal constraints, by proposing the DiStRDF system, which is comprised of a Storage and a Processing Layer. The DiStRDF Storage Layer is responsible for efficiently storing large amount of historical spatio-temporal RDF data of moving objects. On top of it, we devise our DiStRDF Processing Layer, which parses a SPARQL query and produces corresponding logical and physical execution plans. We use Spark, a well-known distributed in-memory processing framework, as the underlying processing engine. Our experimental evaluation, on real data from both aviation and maritime domains, demonstrates the efficiency of our DiStRDF system, when using various spatio-temporal range constraints. |
| 19. | Georgios M. Santipantakis George A. Vouros, Christos Doulkeridis : Coronis: Towards Integrated and Open COVID-19 Data. EDBT 2021, 2021. (Type: Conference | BibTeX) @conference{Santipantakis2021, title = {Coronis: Towards Integrated and Open COVID-19 Data}, author = {Georgios M. Santipantakis, George A. Vouros, Christos Doulkeridis}, year = {2021}, date = {2021-03-26}, booktitle = {EDBT 2021}, keywords = {}, pubstate = {published}, tppubtype = {conference} } |
| 20. | Georgios M. Santipantakis Christos Doulkeridis, George Vouros A: Link Discovery for Maritime Monitoring. In: Springer, 2021, ISBN: 978-3-030-61852-0. (Type: Book Chapter | Abstract | BibTeX) @inbook{Santipantakis2021b, title = {Link Discovery for Maritime Monitoring}, author = {Georgios M. Santipantakis, Christos Doulkeridis, George A. Vouros}, isbn = {978-3-030-61852-0}, year = {2021}, date = {2021-02-09}, publisher = {Springer}, abstract = {Link discovery in the maritime domain is the process of identifying relations—usually of spatial or spatio-temporal nature—between entities that originate from different data sources. Essentially, link discovery is a step towards data integration, which enables interlinking data from disparate sources. As a typical example, vessel trajectories need to be enriched with various types of information: weather conditions, events, contextual data. In turn, this provides enriched data descriptions to data analysis operations, which may lead to the identification of hidden or complex patterns, which would otherwise not be discovered, as they rely on data originating from disparate data sources. This chapter presents the fundamental concepts of link discovery relevant to the maritime domain, focusing on spatial and spatio-temporal data. Due to the processing-intensive nature of the link discovery task over voluminous data, several techniques for efficient processing are presented together with examples on real-world data from the maritime domain.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Link discovery in the maritime domain is the process of identifying relations—usually of spatial or spatio-temporal nature—between entities that originate from different data sources. Essentially, link discovery is a step towards data integration, which enables interlinking data from disparate sources. As a typical example, vessel trajectories need to be enriched with various types of information: weather conditions, events, contextual data. In turn, this provides enriched data descriptions to data analysis operations, which may lead to the identification of hidden or complex patterns, which would otherwise not be discovered, as they rely on data originating from disparate data sources. This chapter presents the fundamental concepts of link discovery relevant to the maritime domain, focusing on spatial and spatio-temporal data. Due to the processing-intensive nature of the link discovery task over voluminous data, several techniques for efficient processing are presented together with examples on real-world data from the maritime domain. |