@conference{Mademlis2023,
title = {Visual inspection for illicit items in x-ray images using deep learning},
author = {Ioannis Mademlis, Georgios Batsis, Adamantia Anna Rebolledo Chrysochoou, Georgios Th Papadopoulos},
url = {https://ieeexplore.ieee.org/abstract/document/10386207/keywords#keywords},
doi = {https://doi.org/10.1109/BigData59044.2023.10386207},
isbn = {979-8-3503-2445-7},
year = {2023},
date = {2023-12-15},
booktitle = {2023 IEEE International Conference on Big Data (BigData)},
pages = {4081-4089},
abstract = {Automated detection of contraband items in X-ray images can significantly increase public safety, by enhancing the productivity and alleviating the mental load of security officers in airports, subways, customs/post offices, etc. The large volume and high throughput of passengers, mailed parcels, etc., during rush hours practically make it a Big Data problem. Modern computer vision algorithms relying on Deep Neural Networks (DNNs) have proven capable of undertaking this task even under resource-constrained and embedded execution scenarios, e.g., as is the case with fast, single-stage object detectors. However, no comparative experimental assessment of the various relevant DNN components/methods has been performed under a common evaluation protocol, which means that reliable cross-method comparisons are missing. This paper presents exactly such a comparative assessment, utilizing a public relevant dataset and a well-defined methodology for selecting the specific DNN components/modules that are being evaluated. The results indicate the superiority of Transformer detectors, the obsolete nature of auxiliary neural modules that have been developed in the past few years for security applications and the efficiency of the CSP-DarkNet backbone CNN.},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}

@conference{Doulkeridis2023,
title = {MobiSpaces: An architecture for energy-efficient data spaces for mobility data},
author = {Christos Doulkeridis, Georgios M Santipantakis, Nikolaos Koutroumanis, George Makridis, Vasilis Koukos, George S Theodoropoulos, Yannis Theodoridis, Dimosthenis Kyriazis, Pavlos Kranas, Diego Burgos, Ricardo Jimenez-Peris, Mariana MG Duarte, Mahmoud Sakr, Esteban Zimányi, Anita Graser, Clemens Heistracher, Kristian Torp, Ioannis Chrysakis, Theofanis Orphanoudakis, Evgenia Kapassa, Marios Touloupou, Jürgen Neises, Petros Petrou, Sophia Karagiorgou, Rosario Catelli, Domenico Messina, Marcelo Corrales Compagnucci, Matteo Falsetta},
url = {https://ieeexplore.ieee.org/abstract/document/10386539},
doi = {https://doi.org/10.1109/BigData59044.2023.10386539},
isbn = {979-8-3503-2445-7},
year = {2023},
date = {2023-12-15},
booktitle = {2023 IEEE International Conference on Big Data (BigData)},
pages = {1487-1494},
abstract = {In this paper, we present an architecture for mobility data spaces enabling trustworthy and reliable data operations along with its main constituent parts. The architecture makes use of a data lake for scalable storage of diverse mobility data sets, on top of which separate computing and storage layers are implemented to allow independent scaling with a data operations toolbox providing all data operations. Furthermore, to cater for mobility analytics, machine learning and artificial intelligence support, an edge analytics suite is provided that encompasses distributed algorithms for mobility analytics and federated learning, thereby exploiting edge computing technologies. In turn, this is supported by a resource allocator that monitors the energy consumption of data-intensive operations and provides this information to the platform for intelligent task placement in edge devices, aiming at energy-efficient operations. As a result, an end-to-end platform is proposed that combines data services and infrastructure services towards supporting mobility application domains, such as urban and maritime.},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}

@conference{Chitos2023,
title = {Nudging households for energy savings via smartphone apps and web portals: An empirical study},
author = {Andreas Chitos, Merkouris Karaliopoulos, Sabine Pelka, Maria Halkidi, Iordanis Koutsopoulos},
url = {https://mm.aueb.gr/publications/197aa2ac-aede-4bd2-8764-edc35cbbd101.pdf},
doi = {https://resolver.tudelft.nl/uuid:b1b7680d-8470-4aa2-905c-df77a16de685},
year = {2023},
date = {2023-11-28},
booktitle = {BEHAVE 2023: 7th European Conference on Behaviour Change for Energy Efficiency},
pages = {293-304},
abstract = {In this paper, we report evidence collected in the context of the Horizon 2020 NUDGE project about the effectiveness of digital tools such as smartphone apps and web portals to realize nudging interventions towards different energy efficiency goals: from the reduction of heating energy and electricity to the increase of self-consumption in energy prosumer households. We analyse recorded events from the interaction of participants with those tools in the context of three different pilot experiments.
We first assess the level of end user engagement with the apps and the portal, counting the number of distinct days that they interact with them. We find it to be highly heterogeneous, with up to 25% of participants in the Greek pilot and 12% in the Portuguese pilot not using the mobile app at all, and the rest forming three distinct groups of low, medium and high engagement. The interaction with the apps almost always lasts fractions of a minute and involves accessing a few app screens. We next turn to the actual users’ exposure to the nudging features of the digital tools to find out that high percentages of users (up to 50%) exhibit zero or very occasional exposure to the app screens that implement nudges. The mobile app users, in particular, can be grouped into four clusters depending on the level of engagement with the app and their exposure to its nudging features. Disappointingly, more than half the pilot participants belong to the cluster combining low engagement with low exposure to nudging. Combining these data with self-statements of participants in postintervention surveys, we find no significant correlation between the level of nudging exposure and the (self-stated) motivation/ intentions to save energy.},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}

@conference{Santipantakis2023b,
title = {An RDF Benchmark for Enriched Maritime Data},
author = {Georgios M Santipantakis, Christos Doulkeridis},
url = {https://dl.acm.org/doi/abs/10.1145/3615885.3628007},
doi = {https://doi.org/10.1145/3615885.3628007},
isbn = {9798400703478},
year = {2023},
date = {2023-11-13},
booktitle = {Proceedings of the 1st ACM SIGSPATIAL International Workshop on Methods for Enriched Mobility Data: Emerging issues and Ethical perspectives 2023},
pages = {1-10},
abstract = {This paper provides an RDF benchmark for triple stores for querying maritime data. The benchmark comes with an integrated data set of real-world data from diverse sources, which has been transformed to RDF triples, and a set of 15 SPARQL queries of varying complexity. Differently from previous available benchmarks that have focused on spatial RDF data, our work targets RDF data about trajectories of moving vessels, therefore the focus is on spatio-temporal RDF data. We present the process to generate the integrated data set, we delve into the details of the benchmark queries, and we provide evaluation results using a selected RDF triple store as a showcase. Furthermore, we make all data, queries and results publicly available to stimulate further research in the field.},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}

@conference{Karakikes2023b,
title = {Towards Handling Bias in Intelligence Analysis with Twitter},
author = {Alexandros Karakikes, Panagiotis Alexiadis, Theocharis Thecharopoulos, Nikolaos Skoulidas, Dimitris Spiliotopoulos, Konstantinos Kotis},
url = {https://ieeexplore.ieee.org/abstract/document/10302618/},
doi = {https://doi.org/10.1109/DSAA60987.2023.10302618},
isbn = {979-8-3503-4503-2},
year = {2023},
date = {2023-11-06},
booktitle = {2023 IEEE 10th International Conference on Data Science and Advanced Analytics (DSAA)},
abstract = {Bias identification and mitigation in the Twitter ecosystem has been lately researched towards achieving a more efficient utilization of the application by different stakeholders and for a wide area of purposes. Among these stakeholders, intelligence services worldwide, collectively called the Intelligence Community (IC), tend to use Twitter, supplementarily to their pre-existent disciplines, for monitoring areas of interest and identifying emerging social, political and security trends/threats. Over time, the IC has identified bias as the major obstacle in information analysis, thus it has developed scientific and empirical methods for bias mitigation, in parallel to those developed by the information and communication technology (ICT) and artificial intelligence (AI) community. As it becomes apparent, it is to both communities’ interest to accurately trace bias and ideally eradicate or moderate its effects. In this paper we draw systemic parallels between Intelligence Analysis (IA) and Twitter Analytics (TA), comparatively examine existing bias mitigating methodologies to pinpoint similarities/dissimilarities, and utterly investigate the feasibility of adapting and adjusting methodologies from the first field to the latter. Furthermore, we propose a novel framework for AI-augmented bias mitigation in the IC. Finally, we propose methods and tools, already adapted by the ICT community, for efficiently supporting bias mitigation methodologies adapted by the IC.},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}

@conference{Zafeiropoulos2023,
title = {Wear4PDmove: an ontology for knowledge-based personalized health monitoring of PD patients},
author = {Nikolaos Zafeiropoulos, Pavlos Bitilis, Konstantinos Kotis},
url = {https://hozo.jp/ISWC2023_PD-Industry-proc/ISWC2023_paper_433.pdf},
year = {2023},
date = {2023-11-01},
booktitle = {The 22nd International Semantic Web Conference (ISWC)},
abstract = {In the field of Parkinson’s Disease (PD), wearable sensors are commonly used to collect movement data from patients for various purposes such as analysis, monitoring, and alerting. To ensure interoperability with other personal health data, such as PHR data, it is crucial to semantically describe this data. Our work focuses on reusing existing ontologies and introducing new conceptualizations to engineer Personal Health Knowledge Graph (PHKG) for PD patient monitoring and doctor alerting. We aim to address the specific knowledge requirements in personal health for PD and support rule-based high-level event recognition. Developing a PHKG can greatly assist health specialists in efficiently assessing patients’ conditions, providing timely and cost-effective care for PD patients.},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}

@conference{Bentos2023,
title = {Bias Assessment in AI-Based Predictions of Recidivism},
author = {S. Bentos, S. Spirou, K. Kotis, G. Tsekouras},
url = {https://metabody.eu/wp-content/uploads/2023/06/13thBHC_ABSTRACT_BOOKLET.pdf},
year = {2023},
date = {2023-09-01},
booktitle = {13th Beyond Humanism Conference (BHC)},
abstract = {Recidivism refers to a person’s relapse into criminal behavior after receiving some form of punishment or undergoes intervention for a previous crime [1, 2]. Numerous individual factors and criminal justice processes (e.g., age, prior arrests, etc.) contribute to the construction of risk assessment instruments. As such, predicting recidivism has significant impact in terms of allocating and managing resources such as in social services, in policy-making decisions, in sentencing planning and probation, in bail options, and in obtaining valuable and prompt insights of the risk posed by various individuals involved in the system [2].
During the last decades, artificial intelligence (AI) algorithms have been used to predict recidivism and guide decisions and choices in managing criminal population by assessing a criminal defendant’s likelihood of committing a crime. Two well-known AI algorithmic frameworks are the COMPAS [3] and the OxRec [4]. Both capture and use certain personal aspects relating to a natural person such as income, marital status, prior alcohol abuse, drug use, and psychological illness of the suspect or alleged offender.
Beyond the adequacy of AI systems in terms of prediction, an important obstacle is the bias that is encoded in the data and/or in the algorithms predicting delinquency or recidivism analysis [1, 5, 6]. Often it has been stated that bias in terms of gender, race and nationality are some of the most sensitive variables that affect the fair decision of AI systems on recidivism of the offenders. These sensitive variables are usually called protected variables. As an example, it has been proved that the COMPAS system obtains biased decisions against black defendants (i.e., the protected variable in this case is race) by classifying them as having twice higher risk of recidivism than white defendants [1, 7].
This study contributes a framework towards assessing the bias related to AI-based recidivism predictions using a set of open-source methods and tools such as the Weka-based machine-learning (ML) algorithms [8], evaluated with a Greek female prison recidivism data set. The data set includes a sample of 6000 females with the following features: (1) year of first release, (2) age of exiting first imprisonment, (3) country of origin, (4) profession before the first imprisonment, (5) education, (6) marital status, (7) number of children. To conduct our experiments, five different ML classification algorithms running on WEKA platform [8] were used: (a) decision tree (J48), (b) naïve Bayes, (c) knearest-neighbors (iBk), (d) logistic regression, and (e) neural network.
By observing and analyzing the classification results of each algorithm, we investigated which attributes associated with re-incarceration are subject to bias. Thus, the goal was to figure out whether AI, in the simplest form of ML, is biased when deciding the fate of a past offender, and ultimately, which algorithm is most effective according to AI models. To obtain our results we quantified the disparate impact [2, 9] of the above seven features. The results showed a significant bias estimation related to the unemployment status before the first imprisonment. In this case, the offender’s profession before the first imprisonment was the protected variable. Future research involves the development of sophisticated algorithms to mitigate and eliminate the bias resulting from the implementation of the above classification algorithms.},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}

@conference{Batsis2023,
title = {Illicit item detection in X-ray images for security applications},
author = {Georgios Batsis, Ioannis Mademlis, Georgios Th Papadopoulos},
url = {https://ieeexplore.ieee.org/abstract/document/10233969},
doi = {https://doi.org/10.1109/BigDataService58306.2023.00016},
isbn = {979-8-3503-3379-4},
year = {2023},
date = {2023-07-17},
booktitle = {2023 IEEE Ninth International Conference on Big Data Computing Service and Applications (BigDataService)},
pages = {63-70},
abstract = {Automated detection of contraband items in X-ray images can significantly increase public safety, by enhancing the productivity and alleviating the mental load of security officers in airports, subways, customs/post offices, etc. The large volume and high throughput of passengers, mailed parcels, etc., during rush hours make it a Big Data analysis task. Modern computer vision algorithms relying on Deep Neural Networks (DNNs) have proven capable of undertaking this task even under resource-constrained and embedded execution scenarios, e.g., as is the case with fast, single-stage, anchor-based object detectors. This paper proposes a two-fold improvement of such algorithms for the X-ray analysis domain, introducing two complementary novelties. Firstly, more efficient anchors are obtained by hierarchical clustering the sizes of the ground-truth training set bounding boxes; thus, the resulting anchors follow a natural hierarchy aligned with the semantic structure of the data. Secondly, the default Non-Maximum Suppression (NMS) algorithm at the end of the object detection pipeline is modified to better handle occluded object detection and to reduce the number of false predictions, by inserting the Efficient Intersection over Union (E-IoU) metric into the Weighted Cluster NMS method. E-IoU provides more discriminative geometrical correlations between the candidate bounding boxes/Regions-of-Interest (RoIs). The proposed method is implemented on a common single-stage object detector (YOLOv5) and its experimental evaluation on a relevant public dataset indicates significant accuracy gains over both the baseline and competing approaches. This highlights the potential of Big Data analysis in enhancing public safety.},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}

@conference{Bitilis2023,
title = {Uncovering the semantics of PD patients’ movement data collected via off-the-shelf wearables},
author = {Pavlos Bitilis, Nikolaos Zafeiropoulos, Adam Koletis, Konstantinos Kotis},
url = {https://ieeexplore.ieee.org/abstract/document/10345958},
doi = {https://doi.org/10.1109/IISA59645.2023.10345958},
year = {2023},
date = {2023-07-10},
booktitle = {The 14th International Conference on Information, Intelligence, Systems and Applications (IISA), Volos, 2023},
abstract = {Wearable sensors are used in monitoring patients with neurodegenerative diseases (ND), such as Parkinson Disease (PD), to collect movement data for the analysis and the assessment of patients’ symptoms. To become interoperable and interlinked with other related personal health data, collected data through sensors embedded in wearable devices need to be semantically described in a commonly agreed, explicit, and formal way. Personal health records (PHRs) including patients’ Magnetic Resonance Imaging (MRIs), medical prescriptions, and medical advice, can provide a unified view of personal health to health specialists, decreasing their efforts to constantly assess patients’ condition via traditional methods. This study aims to present our work for collecting movement data of PD patients through wearables, analyzing them to uncover their inherent semantics, and employing these semantic insights to annotate data in a formal and explicit way to facilitate interlinking with other related heterogeneous data. The movement data was collected via unobstructive wearable technology for health monitoring, and existing formal semantic models were examined for their suitability to be reused and extended for the semantic annotation of the collected movement data. Furthermore, this paper reports early work towards representing such knowledge in the form of a Knowledge Graph (KG) to support rule-based high-level event recognition, such as a missing dose event, for monitoring PD patients and alerting their doctors.},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}

@conference{Karakikes2023,
title = {Understanding bias in Twitter-based intelligence analysis},
author = {Alexandros Karakikes, Panagiotis Alexiadis, Theocharis Theocharopoulos, Nikolaos Skoulidas, Konstantinos Kotis},
url = {https://ieeexplore.ieee.org/abstract/document/10345941},
doi = {https://doi.org/10.1109/IISA59645.2023.10345941},
isbn = {979-8-3503-1806-7},
year = {2023},
date = {2023-07-10},
booktitle = {2023 14th International Conference on Information, Intelligence, Systems & Applications (IISA)},
abstract = {Twitter has been lately engaged by the community of intelligence services worldwide that for monitoring areas of interest and identifying emerging social, political and security trends/threats. Over time, the Intelligence Community (IC) has identified bias as the major obstacle in information analysis, thus it has developed scientific and empirical methods for bias mitigation, in parallel to those developed by the information and communication technology (ICT) and artificial intelligence (AI) community. Both communities share the interest to accurately trace bias and ideally eradicate or moderate its effects. In this paper we introduce systemic parallels between Intelligence Analysis (IA) and Twitter Analytics (TA), to highlight similarities/dissimilarities, and investigate the feasibility of adapting and adjusting methodologies from the first field to the latter. Furthermore, we briefly introduce a novel framework for AI-augmented bias mitigation in the IC, proposing methods and tools, already adapted by the ICT community, for efficiently supporting bias mitigation methodologies adapted by the IC.},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}

@conference{Papadopoulos2023,
title = {Andronicus of rhodes at SemEval-2023 task 4: Transformer-based human value detection using four different neural network architectures},
author = {Georgios Papadopoulos, Marko Kokol, Maria Dagioglou, Georgios Petasis},
url = {https://aclanthology.org/2023.semeval-1.75.pdf},
doi = {https://doi.org/10.18653/v1/2023.semeval-1.75},
year = {2023},
date = {2023-07-01},
booktitle = {Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval-2023)},
pages = {542–548},
publisher = {Association for Computational Linguistics},
abstract = {This paper presents our participation to the “Human Value Detection shared task (Kiesel et al., 2023), as “Andronicus of Rhodes. We describe the approaches behind each entry in the official evaluation, along with the motivation behind each approach. Our best-performing approach has been based on BERT large, with 4 classification heads, implementing two different classification approaches (with different activation and loss functions), and two different partitioning of the training data, to handle class imbalance. Classification is performed through majority voting. The proposed approach outperforms the BERT baseline, ranking in the upper half of the competition.},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}

@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 = {published},
tppubtype = {conference}
}