Can I Trust the Explainer? Verifying Post-hoc Explanatory Methods

TL;DR

This paper critically examines post-hoc explanation methods for neural networks, revealing their limitations and proposing a verification framework based on a real-world neural network architecture to assess explanation reliability.

Contribution

It introduces a verification framework for explanation methods from the feature-selection perspective, addressing current validation shortcomings and exposing explainer failure modes.

Findings

Current explainers often fail under realistic neural network scenarios.

Different explanation perspectives lead to fundamentally different explanations.

The proposed framework provides guarantees on explanation validity.

Abstract

For AI systems to garner widespread public acceptance, we must develop methods capable of explaining the decisions of black-box models such as neural networks. In this work, we identify two issues of current explanatory methods. First, we show that two prevalent perspectives on explanations --- feature-additivity and feature-selection --- lead to fundamentally different instance-wise explanations. In the literature, explainers from different perspectives are currently being directly compared, despite their distinct explanation goals. The second issue is that current post-hoc explainers are either validated under simplistic scenarios (on simple models such as linear regression, or on models trained on syntactic datasets), or, when applied to real-world neural networks, explainers are commonly validated under the assumption that the learned models behave reasonably. However, neural networks often rely on unreasonable correlations, even when producing correct decisions. We introduce a verification framework for explanatory methods under the feature-selection perspective. Our framework is based on a non-trivial neural network architecture trained on a real-world task, and for which we are able to provide guarantees on its inner workings. We validate the efficacy of our evaluation by showing the failure modes of current explainers. We aim for this framework to provide a publicly available, off-the-shelf evaluation when the feature-selection perspective on explanations is needed.