article published Maximilian Muschalik author Fabian Fumagalli author 93420 Barbara Hammer author Eyke Huellermeier author 48129 660 department TRR 318 - C3: TRR 318 - Subproject C3 project TRR 318 - C: TRR 318 - Project Area C project TRR 318: TRR 318 - Erklärbarkeit konstruieren project Explainable Artificial Intelligence (XAI) has mainly focused on static learning tasks so far. In this paper, we consider XAI in the context of online learning in dynamic environments, such as learning from real-time data streams, where models are learned incrementally and continuously adapted over the course of time. More specifically, we motivate the problem of explaining model change, i.e. explaining the difference between models before and after adaptation, instead of the models themselves. In this regard, we provide the first efficient model-agnostic approach to dynamically detecting, quantifying, and explaining significant model changes. Our approach is based on an adaptation of the well-known Permutation Feature Importance (PFI) measure. It includes two hyperparameters that control the sensitivity and directly influence explanation frequency, so that a human user can adjust the method to individual requirements and application needs. We assess and validate our method’s efficacy on illustrative synthetic data streams with three popular model classes. Springer Science and Business Media LLC2022 eng Artificial Intelligence 0933-1875 1610-198710.1007/s13218-022-00766-6 363-4211-224 Muschalik M, Fumagalli F, Hammer B, Huellermeier E. Agnostic Explanation of Model Change based on Feature Importance. <i>KI - Künstliche Intelligenz</i>. 2022;36(3-4):211-224. doi:<a href="https://doi.org/10.1007/s13218-022-00766-6">10.1007/s13218-022-00766-6</a> @article{Muschalik_Fumagalli_Hammer_Huellermeier_2022, title={Agnostic Explanation of Model Change based on Feature Importance}, volume={36}, DOI={<a href="https://doi.org/10.1007/s13218-022-00766-6">10.1007/s13218-022-00766-6</a>}, number={3–4}, journal={KI - Künstliche Intelligenz}, publisher={Springer Science and Business Media LLC}, author={Muschalik, Maximilian and Fumagalli, Fabian and Hammer, Barbara and Huellermeier, Eyke}, year={2022}, pages={211–224} } Muschalik, Maximilian, et al. “Agnostic Explanation of Model Change Based on Feature Importance.” <i>KI - Künstliche Intelligenz</i>, vol. 36, no. 3–4, Springer Science and Business Media LLC, 2022, pp. 211–24, doi:<a href="https://doi.org/10.1007/s13218-022-00766-6">10.1007/s13218-022-00766-6</a>. Muschalik, Maximilian, Fabian Fumagalli, Barbara Hammer, and Eyke Huellermeier. “Agnostic Explanation of Model Change Based on Feature Importance.” <i>KI - Künstliche Intelligenz</i> 36, no. 3–4 (2022): 211–24. <a href="https://doi.org/10.1007/s13218-022-00766-6">https://doi.org/10.1007/s13218-022-00766-6</a>. M. Muschalik, F. Fumagalli, B. Hammer, E. Huellermeier, KI - Künstliche Intelligenz 36 (2022) 211–224. Muschalik, M., Fumagalli, F., Hammer, B., &#38; Huellermeier, E. (2022). Agnostic Explanation of Model Change based on Feature Importance. <i>KI - Künstliche Intelligenz</i>, <i>36</i>(3–4), 211–224. <a href="https://doi.org/10.1007/s13218-022-00766-6">https://doi.org/10.1007/s13218-022-00766-6</a> M. Muschalik, F. Fumagalli, B. Hammer, and E. Huellermeier, “Agnostic Explanation of Model Change based on Feature Importance,” <i>KI - Künstliche Intelligenz</i>, vol. 36, no. 3–4, pp. 211–224, 2022, doi: <a href="https://doi.org/10.1007/s13218-022-00766-6">10.1007/s13218-022-00766-6</a>. 487802023-11-10T14:21:06Z2025-01-16T16:19:35Z