Interpretable Out-Of-Distribution Detection Using Pattern Identification

TL;DR

This paper introduces an interpretable out-of-distribution detection method using pattern identification, which does not require retraining classifiers and provides explanations for confidence scores, demonstrating competitive performance across multiple datasets.

Contribution

The work applies explainable AI techniques to OoD detection, enabling interpretability without retraining classifiers and introducing a new benchmark based on dataset perturbations.

Findings

Detection performance comparable to existing methods.

Pattern identification enhances interpretability of confidence scores.

Robustness depends on classifier architecture and dataset characteristics.

Abstract

Out-of-distribution (OoD) detection for data-based programs is a goal of paramount importance. Common approaches in the literature tend to train detectors requiring inside-of-distribution (in-distribution, or IoD) and OoD validation samples, and/or implement confidence metrics that are often abstract and therefore difficult to interpret. In this work, we propose to use existing work from the field of explainable AI, namely the PARTICUL pattern identification algorithm, in order to build more interpretable and robust OoD detectors for visual classifiers. Crucially, this approach does not require to retrain the classifier and is tuned directly to the IoD dataset, making it applicable to domains where OoD does not have a clear definition. Moreover, pattern identification allows us to provide images from the IoD dataset as reference points to better explain the confidence scores. We demonstrates that the detection capabilities of this approach are on par with existing methods through an extensive benchmark across four datasets and two definitions of OoD. In particular, we introduce a new benchmark based on perturbations of the IoD dataset which provides a known and quantifiable evaluation of the discrepancy between the IoD and OoD datasets that serves as a reference value for the comparison between various OoD detection methods. Our experiments show that the robustness of all metrics under test does not solely depend on the nature of the IoD dataset or the OoD definition, but also on the architecture of the classifier, which stresses the need for thorough experimentations for future work on OoD detection.