Βραβείο καλύτερης δημοσίευσης σε Περιοδικό

• Apostolos Evangelidis – Efficient deep Q-learning for industrial equipment calibration in elevator manufacturingA. Evangelidis, N. Dimitriou, P. Charalampous, T. D. Mastos, D. Tzovaras, “Efficient deep Q-learning for industrial equipment calibration in elevator manufacturing.“, IEEE Transactions on Industrial Informatics, vol. 20, no. 10, pp. 12220 – 12230, October 2024.

Σχόλια επιτροπής αξιολόγησης:

Originality – Excellent originality, devising novel mathematical formulation and subsequently applying to real manufacturing facilities for extensive testing.

Importance – Real world application and real world industrial setting for experimentation.

Rigour – Sound and rigorous methodological approach, supported by extensive experimentation

Βραβείο καλύτερης δημοσίευσης σε Συνέδριο

• Christos Koutlis – Leveraging representations from intermediate encoder-blocks for synthetic image detectionC. Koutlis, S. Papadopoulos, “Leveraging Representations from Intermediate Encoder-blocks for Synthetic Image Detection.“, in Proceedings of the ECCV, 2024.

Σχόλια επιτροπής αξιολόγησης:

Originality – The paper introduces a novel approach to synthetic image detection by leveraging intermediate representations from CLIP’s image encoder. This is an interesting contribution to the field, since prior work mostly focused on final-layer embeddings. The design of the Trainable Importance Estimator adds novelty in the architecture.

Importance – With the exponential rise of generative AI, detecting synthetic media is a real, pressing technical challenge with important societal implications. The method achieves strong generalization across a broad range of generative models, making it a good candidate for real-world deployment in relevant systems, e.g. trust and safety ones. The algorithm outperforms state-of-the-art methods by +10.6% accuracy and trains very fast. It is important to note that the paper has already cited 23 times according to Google Scholar

Rigour – The experimental setup is extensive, covering 20 datasets and including comparisons with state-of-the-art methods, ablation studies, and robustness tests. The model is evaluated in multiple training configurations with clear metric reporting. The level of empirical detail is very good.

Βραβείο καλύτερης δημοσίευσης φοιτητή σε Περιοδικό

• Thanasis Kotsiopoulos – Revolutionizing defect recognition in hard metal industry through AI explainability, human-in-the-loop approaches and cognitive mechanismsT. Kotsiopoulos, G. Papakostas, T. Vafeiadis, V. Dimitriadis, A. Nizamis, A. Bolzoni, D. Bellinati, D. Ioannidis, K. Votis, D. Tzovaras, P. Sarigiannidis, “Revolutionizing defect recognition in hard metal industry through AI explainability, human-in-the-loop approaches and cognitive mechanisms.“, Expert Systems with Applications, vol. 255, December 1, 2024.

Σχόλια επιτροπής αξιολόγησης:

Originality – This paper presents an innovative approach to defect recognition in the hard metal industry by integrating explainable AI (XAI), human-in-the-loop (HITL) techniques, and cognitive retraining mechanisms. Unlike conventional automated defect detection systems that operate as black-box models, this study emphasizes AI transparency and human-AI collaboration. The inclusion of interpretable AI models ensures that AI-driven decisions are understandable to operators, allowing informed interventions and refinements.

Importance – The paper addresses key challenges in industrial defect detection, particularly trust and adaptability in AI systems. Traditional AI-based quality control methods often struggle with operator acceptance and reliability concerns, as their predictions lack intuitive explanations. By incorporating XAI, the platform provides insightful justifications for each AI decision. Furthermore, HITL mechanisms allow real-time expert feedback, enabling models to learn from human corrections and continuously improve their detection accuracy.

Rigour – The research presents a structured methodology, detailing model architecture, image acquisition techniques, and micro-service-based system design. The introduction of retraining mechanisms ensures AI models remain robust over time. The study uses machine and deep learning algorithms for defect classification and localization, leveraging industry-standard AI techniques. However, while simulation results are promising, further experimental validation in real data would be much desirable. Additionally, conducting a comparative analysis against existing AI-based defect detection solutions would strengthen the paper.

Βραβείο καλύτερης δημοσίευσης φοιτητή σε Συνέδριο

• Anestis Kastellos – FedHARM: Harmonizing Model Architectural Diversity in Federated LearningA. Kastellos, A. Psaltis, C. Z. Patrikakis, P. Daras, “FedHARM: Harmonizing Model Architectural Diversity in Federated Learning.“, in Proceedings of the European Conference on Computer Vision (ECCV 2024), Milan, Italy.

Σχόλια επιτροπής αξιολόγησης:

Originality – Unlike standard FL approaches that require homogeneous models for weight aggregation, FedHARM focuses on harmonizing representations rather than model parameters through a hybrid training method.

Importance – The architecture heterogeneity problem appears to be a novel problem that this work handles in FL. As such it is an important contribution to the community although experimentation was performed in simple datasets.

Rigour – The methodology is clearly described, includes implementation details, and presents quantitative evaluations. The experiments are thorough across multiple datasets, architectures, and client counts. The question is how and whether the methodology adapts to other architectures, and different and more modern datasets.

Βραβείο καλύτερης δημοσίευσης σε Περιοδικό

• Kosmas Dimitropoulos – Multi-Manifold Attention for Vision TransformersD. Konstantinidis, I. Papastratis, K. Dimitropoulos, P. Daras, “Multi-manifold attention for vision transformers.“, In IEEE Access, doi: 10.1109/ACCESS.2023.3329952

Σχόλια επιτροπής αξιολόγησης:

Originality – The paper’s originality is in a new self-attention mechanism that uses three different manifolds for the modelling of the input space structure. It introduces the interesting idea of distance maps being computed in each manifold and subsequently fused in an early or late fusion manner.

Importance – The paper has already started attracting attention from the community and its result start becoming influential. The popularity of vision transformers and the fact that the proposed method can be readily applied to any ViT, contribute to its importance and impact.

Rigour – The paper is well written and contains extensive experiments in two tasks (classification and segmentation) that show that the proposed approach improves the performance of ViTs where it is incorporated. Ablation studies are also included.

Βραβείο καλύτερης δημοσίευσης σε Συνέδριο

• Symeon Papadopoulos – Self-Supervised Video Similarity LearningG. Kordopatis-Zilos, G. Tolias, C. Tzelepis, I. Kompatsiaris, I. Patras, S. Papadopoulos, “Self-Supervised Video Similarity Learning.“, 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Vancouver, BC, Canada, 2023, pp. 4756-4766, DOI: 10.1109/CVPRW59228.2023.00504

Σχόλια επιτροπής αξιολόγησης:

Originality – The authors propose a novel self-supervising task for video similarity using data augmentation to learn a video similarity network which then works for several downstream tasks at SoTA without using any labelling data. The approach itself is novel and interesting.

Importance – Self supervision has become a cornerstone of foundation models, models that perform well downstream tasks, and therefore figuring out new ways to perform SSL will dominate the future.

Rigour – The approach is sound, the evaluation proper, and the baselines used appropriate and rigorous.

Βραβείο καλύτερης δημοσίευσης φοιτητή σε Περιοδικό

• Elissavet Batziou – Artistic neural style transfer using CycleGAN and FABEMD by adaptive information selectionE. Batziou, K. Ioannidis, I. Patras, S. Vrochidis, I. Kompatsiaris, “Artistic neural style transfer using CycleGAN and FABEMD by adaptive information selection.“, Pattern Recognition Letters, 165, 55-62. DOI: https://doi.org/10.1016/j.patrec.2022.11.026

Σχόλια επιτροπής αξιολόγησης:

Originality – The presented work uses a cycleGAN adapting its loss function to account for texture information. In order to obtain and transfer the right amount of spectral information the authors define an optimal number of BIMFs, which is the novelty of this work.

Importance – Transfer style is certainly an important and timely area of work, and this work advances the state-of-the-art in the field.

Rigour – The methodology followed in this work is sound. Transfer style evaluation is certainly less rigorous than typical evaluation in ML, even when compared to other generative AI applications, and as such it is always difficult to quantify the benefits of new methods.

Βραβείο καλύτερης δημοσίευσης φοιτητή σε Συνέδριο

• Dimitrios Giakoumis – Leveraging Multimodal Sensing and Topometric Mapping for Human-Like Autonomous Navigation in Complex EnvironmentsK. Tsiakas, D. Alexiou, D. Giakoumis, A. Gasteratos, D. Tzovaras, “Leveraging Multimodal Sensing and Topometric Mapping for Human-Like Autonomous Navigation in Complex Environments“, 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Detroit, MI, USA, 2023, pp. 7415-7421, doi: 10.1109/IROS55552.2023.10341358

Σχόλια επιτροπής αξιολόγησης:

Originality – The proposed method comprises original aspects such as the incorporation of common human driving attitudes in the navigation framework.

Importance – Autonomous driving is an area of immense activity, especially in the robotics and autonomous systems community. The proposed approach contributes to this direction by fusing information from multiple sensors, thus enabling vehicle operation in different, relatively complex environments,

Rigour – There is rigorous experimental evaluation of the work including an experiment involving a real vehicle.

Η διαδικασία συμπεριέλαβε δύο ανεξάρτητες αξιολογήσεις της κάθε υποβληθείσας δημοσίευσης από έναν expert και έναν generalist reviewer σε κλίμακα τεσσάρων κατηγοριών (recognised nationally / recognised internationally / internationally excellent / world-leading), και ακολούθως ranking των εργασιών ανά κατηγορία.

Με στόχο την ανάδειξη των ερευνητικών δραστηριοτήτων και αποτελεσμάτων που διεξάγονται στο ΙΠΤΗΛ, διεξάγεται διαδικασία για την επιλογή καλύτερων δημοσιεύσεων προς βράβευση (Best Paper Awards) στις παρακάτω κατηγορίες.

1. Βραβεία

1. Βραβείο καλύτερης δημοσίευσης σε Περιοδικό
2. Βραβείο καλύτερης δημοσίευσης σε Συνέδριο
3. Βραβείο καλύτερης δημοσίευσης φοιτητή σε Περιοδικό
4. Βραβείο καλύτερης δημοσίευσης φοιτητή σε Συνέδριο

2. Προϋποθέσεις

• Οι δημοσιεύσεις πρέπει να έχουν δημοσιευθεί μέσα στη χρονιά αξιολόγησης σε διεθνή περιοδικά και συνέδρια με κριτές
• Πρέπει να υπάρχει συμφωνία για την υποβολή στη διαδικασία από όλους τους συνσυγγραφείς
• Για όλες τις κατηγορίες η πλειοψηφία των συγγραφέων πρέπει να προέρχονται από το ΙΠΤΗΛ (να έχουν ή είχαν σύμβαση με το ΙΠΤΗΛ)
• Για τις κατηγορίες (3) και (4), ο πρώτος συγγραφέας θα πρέπει να είναι ενεργός μεταπτυχιακός ή διδακτορικός φοιτητής κατά την διάρκεια υποβολής της εργασίας
• Το αντικείμενο της δημοσίευσης πρέπει να ανήκει στον γενικότερο τομέα Πληροφορική και Τηλεπικοινωνιών
• Κάθε εργαστήριο του ΙΠΤΗΛ μπορεί να υποβάλλει μέχρι 3 εργασίες συνολικά για αξιολόγηση για όλες τις κατηγορίες

3. Επιτροπή Αξιολόγησης (2023, 2024)

• Ευάγγελος Κάνουλας, University of Amsterdam
• Σωκράτης Κάτσικας, Norwegian University of Science and Technology
• Νικόλαος Νικολαΐδης, Αριστοτέλειο Πανεπιστήμιο Θεσσαλονίκης
• Ελπινίκη Παπαγεωργίου, Πανεπιστήμιο Θεσσαλίας
• Δημήτριος Πέζαρος, University of Glasgow (Πρόεδρος)

Η διαδικασία περιλαμβάνει δύο ανεξάρτητες αξιολογήσεις της κάθε υποβληθείσας δημοσίευσης από έναν expert και έναν generalist reviewer σε κλίμακα τεσσάρων κατηγοριών (recognised nationally / recognised internationally / internationally excellent / world-leading), και ακολούθως ranking των εργασιών ανά κατηγορία.