Fisher-Rao Sensitivity for Out-of-Distribution Detection in Deep Neural Networks
Abstract
Deep neural networks often remain overconfident on Out-of-Distribution (OoD) inputs. We revisit this problem through Riemannian information geometry. We model the network's predictions as a statistical manifold and find that OoD inputs exhibit higher local Fisher-Rao sensitivity. By quantifying this sensitivity with the trace of the Fisher Information Matrix (FIM), we derive a unifying geometric connection between two common OoD signals: feature magnitude and output uncertainty. Analyzing the limitations of this multiplicative form, we extend our analysis using a product manifold construction. This provides a theoretical framework for the robust additive scores used in state-of-the-art (SOTA) detectors and motivates our final, competitive method.
Cite
Text
Nguyen et al. "Fisher-Rao Sensitivity for Out-of-Distribution Detection in Deep Neural Networks." International Conference on Learning Representations, 2026.Markdown
[Nguyen et al. "Fisher-Rao Sensitivity for Out-of-Distribution Detection in Deep Neural Networks." International Conference on Learning Representations, 2026.](https://mlanthology.org/iclr/2026/nguyen2026iclr-fisherrao/)BibTeX
@inproceedings{nguyen2026iclr-fisherrao,
title = {{Fisher-Rao Sensitivity for Out-of-Distribution Detection in Deep Neural Networks}},
author = {Nguyen, Anthony and Bertrand, Antonin and Le Hégarat-Mascle, Sylvie and Aldea, Emanuel and El-Korso, Mohammed Nabil and Lustrat, Renaud and Florin, Franck},
booktitle = {International Conference on Learning Representations},
year = {2026},
url = {https://mlanthology.org/iclr/2026/nguyen2026iclr-fisherrao/}
}