Playing the network backward: A Game Theoretic Attribution Framework
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917468217278464 |
|---|---|
| author | Zimmermann, Jakob Paul Berend, Jim Loho, Georg Lapuschkin, Sebastian Samek, Wojciech |
| author_facet | Zimmermann, Jakob Paul Berend, Jim Loho, Georg Lapuschkin, Sebastian Samek, Wojciech |
| contents | Attribution methods explain which input features drive a model's prediction, making them central to model debugging and mechanistic interpretability. Yet backward attribution methods, including gradients, LRP, and transformer-specific rules, lack a shared framework in which to compare the underlying backward calculations. We introduce such a framework by recasting backward attribution as a two-player game on an extended network graph, building on Gaubert and Vlassopoulos' ReLU Net Game. Gradients and the full alpha-beta-LRP family arise as integrals over game trajectories under specific equilibria, so attribution maps become projections of trajectory distributions rather than the primary object. Desired explanation properties, such as localisation focus, robustness to input noise, or stable attention routing, can be specified as game-theoretic concepts, including policy regularization, risk aversion, and extended action sets, and translate directly into novel adaptations of the well-known backward rules. On ViT-B/16, one such selected adaptation of alpha-beta-LRP outperforms prior transformer-specific backward methods across all considered localisation metrics. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_06212 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Playing the network backward: A Game Theoretic Attribution Framework Zimmermann, Jakob Paul Berend, Jim Loho, Georg Lapuschkin, Sebastian Samek, Wojciech Machine Learning Computer Vision and Pattern Recognition Attribution methods explain which input features drive a model's prediction, making them central to model debugging and mechanistic interpretability. Yet backward attribution methods, including gradients, LRP, and transformer-specific rules, lack a shared framework in which to compare the underlying backward calculations. We introduce such a framework by recasting backward attribution as a two-player game on an extended network graph, building on Gaubert and Vlassopoulos' ReLU Net Game. Gradients and the full alpha-beta-LRP family arise as integrals over game trajectories under specific equilibria, so attribution maps become projections of trajectory distributions rather than the primary object. Desired explanation properties, such as localisation focus, robustness to input noise, or stable attention routing, can be specified as game-theoretic concepts, including policy regularization, risk aversion, and extended action sets, and translate directly into novel adaptations of the well-known backward rules. On ViT-B/16, one such selected adaptation of alpha-beta-LRP outperforms prior transformer-specific backward methods across all considered localisation metrics. |
| title | Playing the network backward: A Game Theoretic Attribution Framework |
| topic | Machine Learning Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2605.06212 |