Prof. Grzegorz Kołaczek
Wroclaw University of Science and Technology
Poland
grzegorz.kolaczek@pwr.edu.pl

Prof. Krzysztof Brzostowski
Wroclaw University of Science and Technology
Poland
krzysztof.brzostowski@pwr.edu.pl

Dr Jarosław Drapała
Wroclaw University of Science and Technology
Poland
jaroslaw.drapala@pwr.edu.pl

This Special Session focuses on advanced and contemporary machine learning methods that emphasize interpretability, adaptivity, and scalable data-driven learning. We welcome contributions introducing expressive model structures, hybrid optimization strategies, and techniques that improve training, inference, and decision-support in complex real-world settings.
The session welcomes contributions on explainable and structure-aware AI, including models and learning frameworks that offer transparent, verifiable, and human-understandable reasoning. This includes modern interpretable architectures inspired by architectures such as Kolmogorov–Arnold Networks (KAN), compositional or knowledge-guided networks, and symbolic-numeric hybrids. Contributions addressing adaptive and continual learning under distribution shifts, non-stationary environments, or high-dimensional regimes are particularly relevant, together with approaches for integrating and fusing multi-source, heterogeneous, or unstructured data.

The scope of the session includes, but is not limited to, the following topics:

• interpretable and explainable machine learning (XAI),
• structure-aware / knowledge-guided learning architectures (e.g., KAN-inspired,
• compositional models, symbolic-numeric hybrids),
• adaptive, continual, and robust learning under distribution shift or non-stationarity,
• scalable data-driven models for large, heterogeneous, or unstructured datasets,
• data fusion, integration, and multi-source learning,
• hybrid optimization and ML-enhanced decision support,
• applications of ML to complex, distributed, or multi-modal data.