![[2309.13411] Towards Attributions of Input Variables in a Coalition](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2309.13411] Towards Attributions of Input Variables in a Coalition
Summary
This paper addresses the challenge of partitioning input variables in attribution methods for Explainable AI, proposing new metrics to resolve attribution conflicts in coalition scenarios.
Why It Matters
Understanding how to accurately attribute input variables in AI models is crucial for transparency and trust in AI systems. This research provides theoretical insights and practical metrics that enhance the interpretability of AI decisions, which is increasingly important in various applications, from finance to healthcare.
Key Takeaways
- The paper identifies attribution conflicts arising from coalition interactions in AI models.
- It proposes an extension of the Shapley value for improved attribution metrics.
- Three new metrics are introduced to evaluate coalition faithfulness.
- Experiments validate the approach across various domains, including NLP and image classification.
- The findings align with human intuition, enhancing the interpretability of AI systems.
Computer Science > Machine Learning arXiv:2309.13411 (cs) [Submitted on 23 Sep 2023 (v1), last revised 24 Feb 2026 (this version, v3)] Title:Towards Attributions of Input Variables in a Coalition Authors:Xinhao Zheng, Huiqi Deng, Quanshi Zhang View a PDF of the paper titled Towards Attributions of Input Variables in a Coalition, by Xinhao Zheng and 2 other authors View PDF HTML (experimental) Abstract:This paper focuses on the fundamental challenge of partitioning input variables in attribution methods for Explainable AI, particularly in Shapley value-based approaches. Previous methods always compute attributions given a predefined partition but lack theoretical guidance on how to form meaningful variable partitions. We identify that attribution conflicts arise when the attribution of a coalition differs from the sum of its individual variables' attributions. To address this, we analyze the numerical effects of AND-OR interactions in AI models and extend the Shapley value to a new attribution metric for variable coalitions. Our theoretical findings reveal that specific interactions cause attribution conflicts, and we propose three metrics to evaluate coalition faithfulness. Experiments on synthetic data, NLP, image classification, and the game of Go validate our approach, demonstrating consistency with human intuition and practical applicability. Comments: Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV) C...