Efficient and Accurate Explanation Estimation with Distribution Compression

TL;DR

This paper introduces Compress Then Explain (CTE), a novel approach that uses distribution compression to improve the efficiency and accuracy of explanation estimation in machine learning, reducing computational costs significantly.

Contribution

The paper presents a new paradigm, CTE, that leverages kernel thinning for distribution compression, enhancing explanation accuracy with fewer samples and minimal overhead.

Findings

CTE achieves 2-3x faster explanation approximation errors.

CTE reduces the number of model evaluations needed for explanations.

CTE improves stability and accuracy of explanation estimates.

Abstract

We discover a theoretical connection between explanation estimation and distribution compression that significantly improves the approximation of feature attributions, importance, and effects. While the exact computation of various machine learning explanations requires numerous model inferences and becomes impractical, the computational cost of approximation increases with an ever-increasing size of data and model parameters. We show that the standard i.i.d. sampling used in a broad spectrum of algorithms for post-hoc explanation leads to an approximation error worthy of improvement. To this end, we introduce Compress Then Explain (CTE), a new paradigm of sample-efficient explainability. It relies on distribution compression through kernel thinning to obtain a data sample that best approximates its marginal distribution. CTE significantly improves the accuracy and stability of explanation estimation with negligible computational overhead. It often achieves an on-par explanation approximation error 2-3x faster by using fewer samples, i.e. requiring 2-3x fewer model evaluations. CTE is a simple, yet powerful, plug-in for any explanation method that now relies on i.i.d. sampling.