A Bayesian Nonparametric Perspective on Mahalanobis Distance for Out of Distribution Detection

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Linderman, Randolph W., Cowan, Noah, Chen, Yiran, Linderman, Scott W.
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916051290161152
author Linderman, Randolph W.
Cowan, Noah
Chen, Yiran
Linderman, Scott W.
author_facet Linderman, Randolph W.
Cowan, Noah
Chen, Yiran
Linderman, Scott W.
contents Bayesian nonparametric methods are naturally suited to the problem of out-of-distribution (OOD) detection. However, these techniques have largely been eschewed in favor of simpler methods based on distances between pre-trained or learned embeddings of data points. Here we show a formal relationship between Bayesian nonparametric models and the relative Mahalanobis distance score (RMDS), a commonly used method for OOD detection. Building on this connection, we propose Bayesian nonparametric mixture models with hierarchical priors that generalize the RMDS. We evaluate these models on the OpenOOD detection benchmark and show that Bayesian nonparametric methods can improve upon existing OOD methods, especially in regimes where training classes differ in their covariance structure and where there are relatively few data points per class.
format Preprint
id arxiv_https___arxiv_org_abs_2502_08695
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Bayesian Nonparametric Perspective on Mahalanobis Distance for Out of Distribution Detection
Linderman, Randolph W.
Cowan, Noah
Chen, Yiran
Linderman, Scott W.
Machine Learning
Bayesian nonparametric methods are naturally suited to the problem of out-of-distribution (OOD) detection. However, these techniques have largely been eschewed in favor of simpler methods based on distances between pre-trained or learned embeddings of data points. Here we show a formal relationship between Bayesian nonparametric models and the relative Mahalanobis distance score (RMDS), a commonly used method for OOD detection. Building on this connection, we propose Bayesian nonparametric mixture models with hierarchical priors that generalize the RMDS. We evaluate these models on the OpenOOD detection benchmark and show that Bayesian nonparametric methods can improve upon existing OOD methods, especially in regimes where training classes differ in their covariance structure and where there are relatively few data points per class.
title A Bayesian Nonparametric Perspective on Mahalanobis Distance for Out of Distribution Detection
topic Machine Learning
url https://arxiv.org/abs/2502.08695