Provably Better Explanations with Optimized Aggregation of Feature Attributions
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866913381557993472 |
|---|---|
| author | Decker, Thomas Bhattarai, Ananta R. Gu, Jindong Tresp, Volker Buettner, Florian |
| author_facet | Decker, Thomas Bhattarai, Ananta R. Gu, Jindong Tresp, Volker Buettner, Florian |
| contents | Using feature attributions for post-hoc explanations is a common practice to understand and verify the predictions of opaque machine learning models. Despite the numerous techniques available, individual methods often produce inconsistent and unstable results, putting their overall reliability into question. In this work, we aim to systematically improve the quality of feature attributions by combining multiple explanations across distinct methods or their variations. For this purpose, we propose a novel approach to derive optimal convex combinations of feature attributions that yield provable improvements of desired quality criteria such as robustness or faithfulness to the model behavior. Through extensive experiments involving various model architectures and popular feature attribution techniques, we demonstrate that our combination strategy consistently outperforms individual methods and existing baselines. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_05090 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Provably Better Explanations with Optimized Aggregation of Feature Attributions Decker, Thomas Bhattarai, Ananta R. Gu, Jindong Tresp, Volker Buettner, Florian Machine Learning Artificial Intelligence Computer Vision and Pattern Recognition Using feature attributions for post-hoc explanations is a common practice to understand and verify the predictions of opaque machine learning models. Despite the numerous techniques available, individual methods often produce inconsistent and unstable results, putting their overall reliability into question. In this work, we aim to systematically improve the quality of feature attributions by combining multiple explanations across distinct methods or their variations. For this purpose, we propose a novel approach to derive optimal convex combinations of feature attributions that yield provable improvements of desired quality criteria such as robustness or faithfulness to the model behavior. Through extensive experiments involving various model architectures and popular feature attribution techniques, we demonstrate that our combination strategy consistently outperforms individual methods and existing baselines. |
| title | Provably Better Explanations with Optimized Aggregation of Feature Attributions |
| topic | Machine Learning Artificial Intelligence Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2406.05090 |