Mahalanobis++: Improving OOD Detection via Feature Normalization

Published in ICML, 2025

Recommended citation: M. Müller and Matthias Hein (2025) “Mahalanobis++: Improving OOD Detection via Feature Normalization” ICML 2025 https://arxiv.org/abs/2505.18032v1

We show that basic assumptions underlying the Mahalanobis Distance for OOD detection are often severely violated, but can be improved by normalizing the features. The resulting method, Mahalanobis++, improves OOD detection consistently across architectures and pretraining schemes, and leads to new SOTA results.

Download paper here

Recommended citation: M. Müller and Matthias Hein (2025) “Mahalanobis++: Improving OOD Detection via Feature Normalization” ICML 2025