A Dual-Perspective Approach to Evaluating Feature Attribution Methods

TL;DR

This paper introduces a dual-perspective framework for evaluating feature attribution methods in neural networks, focusing on faithfulness, soundness, and completeness to improve assessment accuracy.

Contribution

It proposes two new, mathematically grounded perspectives—soundness and completeness—for evaluating the quality of feature attributions.

Findings

New metrics for soundness and completeness introduced

Applied metrics to mainstream attribution methods

Enhanced understanding of attribution method effectiveness

Abstract

Feature attribution methods attempt to explain neural network predictions by identifying relevant features. However, establishing a cohesive framework for assessing feature attribution remains a challenge. There are several views through which we can evaluate attributions. One principal lens is to observe the effect of perturbing attributed features on the model's behavior (i.e., faithfulness). While providing useful insights, existing faithfulness evaluations suffer from shortcomings that we reveal in this paper. In this work, we propose two new perspectives within the faithfulness paradigm that reveal intuitive properties: soundness and completeness. Soundness assesses the degree to which attributed features are truly predictive features, while completeness examines how well the resulting attribution reveals all the predictive features. The two perspectives are based on a firm mathematical foundation and provide quantitative metrics that are computable through efficient algorithms. We apply these metrics to mainstream attribution methods, offering a novel lens through which to analyze and compare feature attribution methods.