Feature Space Singularity for Out-of-Distribution Detection

TL;DR

This paper introduces a novel OoD detection method based on the concentration of out-of-distribution features in the neural network's feature space, achieving state-of-the-art results and robustness.

Contribution

The paper proposes the Feature Space Singularity (FSS) method, a simple algorithm that detects OoD samples by measuring their distance to a learned feature space center, with theoretical analysis and practical robustness.

Findings

Achieves state-of-the-art OoD detection performance

Demonstrates robustness to test data corruption

Enables further improvement through ensembling

Abstract

Out-of-Distribution (OoD) detection is important for building safe artificial intelligence systems. However, current OoD detection methods still cannot meet the performance requirements for practical deployment. In this paper, we propose a simple yet effective algorithm based on a novel observation: in a trained neural network, OoD samples with bounded norms well concentrate in the feature space. We call the center of OoD features the Feature Space Singularity (FSS), and denote the distance of a sample feature to FSS as FSSD. Then, OoD samples can be identified by taking a threshold on the FSSD. Our analysis of the phenomenon reveals why our algorithm works. We demonstrate that our algorithm achieves state-of-the-art performance on various OoD detection benchmarks. Besides, FSSD also enjoys robustness to slight corruption in test data and can be further enhanced by ensembling. These make FSSD a promising algorithm to be employed in real world. We release our code at \url{https://github.com/megvii-research/FSSD_OoD_Detection}.