Approximating the Shapley Value without Marginal Contributions

TL;DR

This paper introduces two new approximation algorithms for the Shapley value that do not rely on marginal contributions, offering strong theoretical guarantees and empirical validation in explainability contexts.

Contribution

It presents the first domain-independent, parameter-free algorithms for approximating the Shapley value without using marginal contributions, with proven theoretical bounds.

Findings

The algorithms achieve superior approximation quality compared to existing methods.

Empirical results demonstrate effectiveness in synthetic and real explainability tasks.

The methods outperform state-of-the-art approaches in accuracy and efficiency.

Abstract

The Shapley value, which is arguably the most popular approach for assigning a meaningful contribution value to players in a cooperative game, has recently been used intensively in explainable artificial intelligence. Its meaningfulness is due to axiomatic properties that only the Shapley value satisfies, which, however, comes at the expense of an exact computation growing exponentially with the number of agents. Accordingly, a number of works are devoted to the efficient approximation of the Shapley value, most of them revolve around the notion of an agent's marginal contribution. In this paper, we propose with SVARM and Stratified SVARM two parameter-free and domain-independent approximation algorithms based on a representation of the Shapley value detached from the notion of marginal contribution. We prove unmatched theoretical guarantees regarding their approximation quality and provide empirical results including synthetic games as well as common explainability use cases comparing ourselves with state-of-the-art methods.