Probabilistic Stability Guarantees for Feature Attributions

Helen Jin, Anton Xue, Weiqiu You, Surbhi Goel, Eric Wong

Keywords: Explainability, Interpretability, Feature Attributions, Safety, Robustness

TL;DR: We scale certified explanations to provide practical guarantees in high-dimensional settings.

Abstract: Stability guarantees have emerged as a principled way to evaluate feature attributions, but existing certification methods rely on heavily smoothed classifiers and often produce conservative guarantees. To address these limitations, we introduce soft stability and propose a simple, model-agnostic, sample-efficient stability certification algorithm (SCA) that yields non-trivial and interpretable guarantees for any attribution method. Moreover, we show that mild smoothing achieves a more favorable trade-off between accuracy and stability, avoiding the aggressive compromises made in prior certification methods. To explain this behavior, we use Boolean function analysis to derive a novel characterization of stability under smoothing. We evaluate SCA on vision and language tasks and demonstrate the effectiveness of soft stability in measuring the robustness of explanation methods.

Primary Area: Social and economic aspects of machine learning (e.g., fairness, interpretability, human-AI interaction, privacy, safety, strategic behavior)

Submission Number: 16083