Towards Attributions of Input Variables in a Coalition
This addresses a fundamental challenge in Explainable AI for researchers and practitioners, though it appears incremental as it builds on existing Shapley value frameworks.
The paper tackles the problem of partitioning input variables in Shapley value-based attribution methods for Explainable AI, where previous methods lacked theoretical guidance. It proposes a new attribution metric for variable coalitions, validated through experiments on synthetic data, NLP, image classification, and Go, showing consistency with human intuition.
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.