Concept-based Explanations for Out-Of-Distribution Detectors

TL;DR

This paper introduces a method to interpret out-of-distribution detectors using high-level concepts, proposing new metrics and an unsupervised framework to improve explanation quality and understanding of detector decisions.

Contribution

It presents novel metrics for evaluating concept explanations and an unsupervised approach to learn concepts that enhance interpretability of OOD detectors.

Findings

Metrics effectively assess explanation quality.

Framework improves concept-based explanations.

Enhanced understanding of OOD detector decisions.

Abstract

Out-of-distribution (OOD) detection plays a crucial role in ensuring the safe deployment of deep neural network (DNN) classifiers. While a myriad of methods have focused on improving the performance of OOD detectors, a critical gap remains in interpreting their decisions. We help bridge this gap by providing explanations for OOD detectors based on learned high-level concepts. We first propose two new metrics for assessing the effectiveness of a particular set of concepts for explaining OOD detectors: 1) detection completeness, which quantifies the sufficiency of concepts for explaining an OOD-detector's decisions, and 2) concept separability, which captures the distributional separation between in-distribution and OOD data in the concept space. Based on these metrics, we propose an unsupervised framework for learning a set of concepts that satisfy the desired properties of high detection completeness and concept separability, and demonstrate its effectiveness in providing concept-based explanations for diverse off-the-shelf OOD detectors. We also show how to identify prominent concepts contributing to the detection results, and provide further reasoning about their decisions.