Stabilizing Estimates of Shapley Values with Control Variates

TL;DR

This paper introduces ControlSHAP, a Monte Carlo-based method that significantly reduces the variability in Shapley value estimates for model explanations without extra computational cost.

Contribution

It presents a novel control variates approach for stabilizing Shapley value estimates applicable to any machine learning model.

Findings

Reduces Monte Carlo variability in Shapley estimates

Applicable to high-dimensional datasets

Requires minimal additional computation

Abstract

Shapley values are among the most popular tools for explaining predictions of blackbox machine learning models. However, their high computational cost motivates the use of sampling approximations, inducing a considerable degree of uncertainty. To stabilize these model explanations, we propose ControlSHAP, an approach based on the Monte Carlo technique of control variates. Our methodology is applicable to any machine learning model and requires virtually no extra computation or modeling effort. On several high-dimensional datasets, we find it can produce dramatic reductions in the Monte Carlo variability of Shapley estimates.