Towards Attributions of Input Variables in a Coalition

Xinhao Zheng, Huiqi Deng, Quanshi Zhang
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:78115-78138, 2025.

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.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-zheng25d, title = {Towards Attributions of Input Variables in a Coalition}, author = {Zheng, Xinhao and Deng, Huiqi and Zhang, Quanshi}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {78115--78138}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/zheng25d/zheng25d.pdf}, url = {https://proceedings.mlr.press/v267/zheng25d.html}, 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.} }
Endnote
%0 Conference Paper %T Towards Attributions of Input Variables in a Coalition %A Xinhao Zheng %A Huiqi Deng %A Quanshi Zhang %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-zheng25d %I PMLR %P 78115--78138 %U https://proceedings.mlr.press/v267/zheng25d.html %V 267 %X 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.
APA
Zheng, X., Deng, H. & Zhang, Q.. (2025). Towards Attributions of Input Variables in a Coalition. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:78115-78138 Available from https://proceedings.mlr.press/v267/zheng25d.html.

Related Material