CADet: Fully Self-Supervised Out-Of-Distribution Detection With Contrastive Learning

Published: 21 Sept 2023, Last Modified: 02 Nov 2023NeurIPS 2023 posterEveryoneRevisionsBibTeX
Keywords: Contrastive learning, OOD detection, adversarial detection, MMD, ImageNet-O, Anomaly detection, CIFAR-10.1
TL;DR: We leverage self-supervised contrastive learning first with MMD two-sample test to discriminate between two distribution, then in a novel method called CADet for detection of both novel classes and adversarial attacks on single samples.
Abstract: Handling out-of-distribution (OOD) samples has become a major stake in the real-world deployment of machine learning systems. This work explores the use of self-supervised contrastive learning to the simultaneous detection of two types of OOD samples: unseen classes and adversarial perturbations. First, we pair self-supervised contrastive learning with the maximum mean discrepancy (MMD) two-sample test. This approach enables us to robustly test whether two independent sets of samples originate from the same distribution, and we demonstrate its effectiveness by discriminating between CIFAR-10 and CIFAR-10.1 with higher confidence than previous work. Motivated by this success, we introduce CADet (Contrastive Anomaly Detection), a novel method for OOD detection of single samples. CADet draws inspiration from MMD, but leverages the similarity between contrastive transformations of a same sample. CADet outperforms existing adversarial detection methods in identifying adversarially perturbed samples on ImageNet and achieves comparable performance to unseen label detection methods on two challenging benchmarks: ImageNet-O and iNaturalist. Significantly, CADet is fully self-supervised and requires neither labels for in-distribution samples nor access to OOD examples.
Supplementary Material: pdf
Submission Number: 9170