On the detection of Out-Of-Distribution samples in Multiple Instance Learning

TL;DR

This paper explores out-of-distribution detection in Multiple Instance Learning, adapting existing methods, introducing a new benchmark, and analyzing their performance across diverse datasets to highlight challenges and future directions.

Contribution

It adapts post-hoc OOD detection methods to MIL, introduces a dedicated benchmark, and provides extensive experimental analysis highlighting current limitations.

Findings

KNN performs best overall but has notable shortcomings.

The complexity of OOD detection in MIL is high, requiring more robust methods.

The new benchmark facilitates future research in weakly supervised OOD detection.

Abstract

The deployment of machine learning solutions in real-world scenarios often involves addressing the challenge of out-of-distribution (OOD) detection. While significant efforts have been devoted to OOD detection in classical supervised settings, the context of weakly supervised learning, particularly the Multiple Instance Learning (MIL) framework, remains under-explored. In this study, we tackle this challenge by adapting post-hoc OOD detection methods to the MIL setting while introducing a novel benchmark specifically designed to assess OOD detection performance in weakly supervised scenarios. Across extensive experiments based on diverse public datasets, KNN emerges as the best-performing method overall. However, it exhibits significant shortcomings on some datasets, emphasizing the complexity of this under-explored and challenging topic. Our findings shed light on the complex nature of OOD detection under the MIL framework, emphasizing the importance of developing novel, robust, and reliable methods that can generalize effectively in a weakly supervised context. The code for the paper is available here: https://github.com/loic-lb/OOD_MIL.