ExplAI 2025
Advanced AI in Explainability and Ethics for the Sustainable Development Goals

Table of Contents

• Preface
  Summary: There were 44 papers submitted for peer-review to this workshop from authors representing 8 countries. Each submission underwent a rigorous double-blind peer-review process by at least two independent PC members. Out of these, 15 papers were accepted for this volume, resulting in an acceptance rate of approximately 34%.

Session 1: Explainable AI

• Explainable artificial intelligence foundations for web-based sea ice extent forecasting system
  Tetiana Hovorushchenko, Olga Pavlova, Vitalii Alekseiko, Oleg Voichur, Valeriia Shvaiko, Artem Boyarchuk
• CNNs are explainable domain-specific visual embedders
  Zakhar Ostrovsky, Andrii Biloshchytskyi, Dmytro Uhryn
• Neural network detection of digital fatigue and burnout with interpretable thematic segmentation
  Olexander Mazurets, Roman Vit, Maryna Molchanova, Olena Sobko, Adam Wierzbicki, Dmytro Chumachenko
• An explainable artificial intelligence approach for detecting network attacks
  Dmytro Tymoshchuk, Andriy Sverstiuk, Yurii Klots, Nataliia Petliak, Vira Titova

Session 2: AI Ethics and Governance

• A method for synthesising system architecture for IT infrastructure resistant to social engineering attacks
  Sergii Lysenko, Oleksandr Bokhonko, Tomas Sochor, Olha Atamaniuk, Nadiia Lysenko, Jiri Balej
• Linear ensemble model with winner-takes-all aggregation strategy for improved small data classification
  Ivan Izonin, Roman Tkachenko, Serhii Chesanov, Yaroslav Tolstyak, Myroslav Stupnytskyi
• An approach to identifying functional and non-functional requirements in IT-project management using deep learning models
  Ruslan Bahrii, Tetiana Skrypnyk, Bohdan Romanov, Elena Zaitseva, Houda El Bouhissi, Volodymyr Lytvynenko

Session 3: Sustainable Development Goals

• AutoML PyCaret and SHAP explainable AI for ECG signal classification based on amplitude variability
  Dmytro Tymoshchuk, Iryna Didych, Andriy Sverstiuk, Lyubomyr Mosiy, Yuri Palianytsia
• A method for detecting intestinal thrombosis based on a hybrid approach using grey wolf optimization and a genetic algorithm
  Lyubomyr Chyrun, Dmytro Uhryn, Yuriy Ushenko, Oleksii Iliuk, Yuriy Masikevych
• A benchmark methodology for urban traffic pattern clustering using SUMO-based expert-verified ground truth
  Vitaliy Pavlyshyn, Eduard Manziuk, Adnène Arbi, Nebojsa Bacanin, Iurii Krak
• Knowledge engineering information technology for cultural-educational scenarios based on RAG
  Khrystyna Lipianina-Honcharenko, Nazar Melnyk, Myroslav Komar, Pavlo Bykovyy, Khrystyna Yurkiv

Session 4: Trustworthy AI

• Vision transformers transfer learning for smoking detection in public spaces with transparent AI decisions
  Olexander Mazurets, Maryna Molchanova, Olha Zalutska, Ihor Kok, Vitaly Levashenko, Abdel-Badeeh M. Salem
• An approach to matching satellite and UAV images for visual place recognition using color normalization and YOLO
  Volodymyr Vozniak, Yuriy Ushenko, Orken Mamyrbayev
• Verifiable by construction: Evidence-anchored LLMs for explainable fake news detection
  Andrii Shupta, Pavlo Radiuk, Miroslav Kvassay, Piotr Gaj
• Hierarchical neural network model for identifying similar objects in drone images
  Dmytro Borovyk, Oleksander Barmak, Pawel Komada, Sergii Babichev