GEPC: Group-Equivariant Posterior Consistency for Out-of-Distribution Detection in Diffusion Models

Yadang Alexis Rouzoumka, Jean Pinsolle, Eugénie Terreaux, Christèle Morisseau, Jean-Philippe Ovarlez, Chengfang Ren · 2026

Read GEPC: Group-Equivariant Posterior Consistency for Out-of-Distribution Detection in Diffusion Models by Yadang Alexis Rouzoumka, Jean Pinsolle, Eugénie Terreaux, Christèle Morisseau, Jean-Philippe Ovarlez, Chengfang Ren (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-00191.txt.

Description

Yadang Alexis Rouzoumka and coauthors introduce GEPC, a training free probe for out of distribution detection in diffusion models. Diffusion scores often inherit near symmetries such as flips, rotations, and circular shifts from training data and convolutional nets. Most detectors use score size or local geometry and ignore those symmetries. GEPC measures, for a finite group, the consistency with which the learned score transforms, catching symmetry breaking even when score size stays flat. It needs only score checks, gives readable breaking maps, and shows gains on image benchmarks.

Catalog record

CREATORYadang Alexis Rouzoumka, Jean Pinsolle, Eugénie Terreaux, Christèle Morisseau, Jean-Philippe Ovarlez, Chengfang Ren
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE91305
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602.00191

Browse the full library · About the Tower of Babel