arXiv：cs.LG· Jakob Paul Zimmermann, Jim Berend, Georg Loho, Sebastian Lapuschkin, Wojciech Samek·· 13 小时前
把网络反向运行：一个博弈论归因框架
Playing the network backward: A Game Theoretic Attribution Framework
AI 导读
研究者提出一个博弈论框架，将梯度、LRP 及 Transformer 专用规则等反向归因方法统一重述为扩展网络图上的双人博弈，归因图由此成为博弈轨迹分布的投影。
正文
Abstract: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.
| Comments: | 79 pages (9 pages main text, 66 pages appendix). Code: this https URL |
| Subjects: | Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2605.06212 [cs.LG] |
| (or arXiv:2605.06212v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2605.06212 arXiv-issued DOI via DataCite |
Submission history
From: Jakob Paul Zimmermann [view email]
[v1]
Thu, 7 May 2026 13:15:07 UTC (6,352 KB)
[v2]
Fri, 9 Oct 2026 09:29:16 UTC (5,564 KB)
来源：arXiv：cs.LG · arxiv.org