mgr inż. Dorota Gil
ORCID: 0000-0001-6663-7571
The topic of my dissertation is: „Federated learning based on interval-valued fuzzy systems“
Real-world data are rarely precise: they are affected by noise, missing values and differences between sources. Interval-valued fuzzy sets (IVFS) represent such uncertainty explicitly, and fuzzy inference systems built on them provide interpretable, rule-based models. My research concerns federated learning in which local models are interval-valued fuzzy rule-based systems. Data stay at their source, and only rule bases are exchanged and combined. The main problems are the aggregation of rule bases from many clients, the formalization and measurement of conflicts between rules, and the fusion of knowledge while preserving uncertainty and interpretability.
The theoretical objective is a formal framework for fusing interval-valued fuzzy knowledge in a distributed setting. The practical objective is an open-source implementation, extending the IFIS Python library, with experimental verification on real data (e.g., fall detection, sports injury prediction). The dissertation aims to combine privacy-preserving distributed learning with transparent, uncertainty-aware inference, which has so far been studied mainly in a centralized setting.
Publication:
1. Pękala B., Szkoła J., Grochowalski P., Gil D., Kosior D., Dyczkowski K., A Novel Method for Human Fall Detection Using Federated Learning and Interval-Valued Fuzzy Inference Systems, Journal of Artificial Intelligence and Soft Computing Research, 2025, Vol. 15, No. 1, pp. 77–90.
2. Dyczkowski K., Grochowalski P., Kosior D., Gil D., Kozioł W., Pękala B., Kaymak U., Fuchs C., Nobile M.S., Python library for interval-valued fuzzy inference, SoftwareX, 2024, Vol. 26, 101730.
3. Mroczek T., Gil D., Pękala B., Fuzzy and rough approach to the problem of missing data in fall detection system, Fuzzy Sets and Systems, 2024, Vol. 480, 108868.
4. Gil D., Pękala B., The Hybrid Method of Inference System Based on Experts’ Rules and Machine Learning with an Uncertainty Aspect, IPMU 2024 (Part III), Springer, pp. 289–300.
5. Pękala B., Grochowalski P., Kosior D., Gil D., Kozioł W., Dyczkowski K., Kaymak U., Fuchs C., Nobile M.S., Applications of IFIS python library in interval-valued fuzzy inference problems, 2024 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE), pp. 1–9.
6. Pękala B., Mroczek T., Gil D., Kępski M., Application of Fuzzy and Rough Logic to Posture Recognition in Fall Detection System, Sensors, 2022, Vol. 22, No. 4, 1602.
7. Pękala B., Drygaś P., Knap M., Gil D., Kwiatkowski B., Preference and weak interval-valued operator in decision making problem, 2021 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE), pp. 1–7.
8. Gil D., Grochowina M., Leniowska L., Detecting of the rolling bearing state based on acoustic signal and the k-NN classifier, Signal Processing: Algorithms, Architectures, Arrangements, and Applications (SPA 2019), Poznań, IEEE, pp. 246–249.