A Simple Fix to Mahalanobis Distance for Improving Near-OOD Detection
2021-06-16Code Available1· sign in to hype
Jie Ren, Stanislav Fort, Jeremiah Liu, Abhijit Guha Roy, Shreyas Padhy, Balaji Lakshminarayanan
Code Available — Be the first to reproduce this paper.
ReproduceCode
- github.com/google/uncertainty-baselinesOfficialIn papertf★ 1,569
- github.com/stanislavfort/exploring_the_limits_of_OOD_detectionjax★ 44
- github.com/stanislavfort/adversaries_to_ood_detectionjax★ 13
- github.com/glhr/ood-labelnoisepytorch★ 9
Abstract
Mahalanobis distance (MD) is a simple and popular post-processing method for detecting out-of-distribution (OOD) inputs in neural networks. We analyze its failure modes for near-OOD detection and propose a simple fix called relative Mahalanobis distance (RMD) which improves performance and is more robust to hyperparameter choice. On a wide selection of challenging vision, language, and biology OOD benchmarks (CIFAR-100 vs CIFAR-10, CLINC OOD intent detection, Genomics OOD), we show that RMD meaningfully improves upon MD performance (by up to 15% AUROC on genomics OOD).