Metraipolis › archive › dc_c56534ddb678
Rank-1 Weight-Corrected Trainable Weight Averaging for Out-of-Distribution Detection — arXiv:2610.33909
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.33909
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-27T08:00:35.124Z
- SHA-256 of the archived text
db761ea52dcd090c0532df8457387cc71b48f65781c82a908de5f725e9ad89be- Stable URL of this page
https://metraipolis.com/documento/dc_c56534ddb678
How to cite
Rank-1 Weight-Corrected Trainable Weight Averaging for Out-of-Distribution Detection — arXiv:2610.33909 — arXiv (en), https://arxiv.org/abs/2610.33909 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_c56534ddb678 (Metraipolis archive dc_c56534ddb678, sha256 db761ea52dcd090c0532df8457387cc71b48f65781c82a908de5f725e9ad89be)
BibTeX
@misc{dcc56534ddb678,
title = {Rank-1 Weight-Corrected Trainable Weight Averaging for Out-of-Distribution Detection — arXiv:2610.33909},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.33909},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_c56534ddb678; sha256 db761ea52dcd090c0532df8457387cc71b48f65781c82a908de5f725e9ad89be},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.33909. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Out-of-distribution (OOD) detection identifies inputs outside the training distribution. Rank-based analyses such as RankFeat and RankWeight suggest that OOD overconfidence is closely related to dominant low-rank components in feature and weight spaces. Since weight averaging operates directly in parameter space and improves model generalization, we consider it for OOD detection from a dominant-component perspective. Trainable Weight Averaging (TWA) assigns a single coefficient to each checkpoint, limiting independent control of dominant weight components. We extend TWA with leading rank-one weight components extracted by Singular Value Decomposition as correction directions, whose contributions are learned through the in-distribution classification objective. On ResNet50, our method improves over fine-tuning and standard TWA across the evaluated datasets with MSP and RankWeight, both with and without Outlier Exposure. Mahalanobis-based evaluation yields mixed results, indicating that the benefit depends on the OOD scoring protocol. Target-layer analyses with MSP and Energy favor correction near the classifier. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI); Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2610.33909 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-27T07:55:30.863Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).