L2-normalizing pre-logit features improves Mahalanobis-based out-of-distribution detection across 44 ImageNet models, reducing the average false positive rate by 7.6 percentage points.
A framework for benchmarking class-out-of-distribution detection and its application to imagenet
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Mahalanobis++: Improving OOD Detection via Feature Normalization
L2-normalizing pre-logit features improves Mahalanobis-based out-of-distribution detection across 44 ImageNet models, reducing the average false positive rate by 7.6 percentage points.