Shapley Value Approximation based on k-Additive Games
Keywords: Explainable AI, Shapley Value, Feature Importance, Game Theory
TL;DR: We propose a new method using k-additive games to estimate Shapley values commonly used in Explainable AI
Abstract: The Shapley value is the prevalent solution for fair division problems in which a payout is to be divided among multiple agents. By adopting a game-theoretic view, the idea of fair division and the Shapley value can also be used in machine learning to quantify the individual contribution of features or data points to the performance of a predictive model. Despite its popularity and axiomatic justification, the Shapley value suffers from a computational complexity that scales exponentially with the number of entities involved, and hence requires approximation methods for its reliable estimation. In this paper, we propose SVA$k_{\text{ADD}}$, a novel approximation method that fits a $k$-additive surrogate game. By taking advantage of the assumption of $k$-additivity, we are able to compute the exact Shapley values of the surrogate game in polynomial time, and then use these values as estimates for the original fair division problem. The efficacy of our method is evaluated empirically and compared to competing methods.
Primary Area: interpretability and explainable AI
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 10314
Loading