Metraipolis › archive › dc_9cd242e49ebe
Covariance-aware Scoring: A Geometric Correction for Vision-Language Test-Time Adaptation — arXiv:2611.00024
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.00024
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T11:23:28.534Z
- SHA-256 of the archived text
4d415dd536031052d70f658b2521fac67c7b553322e7a24ddd2d191dcce537a8- Stable URL of this page
https://metraipolis.com/documento/dc_9cd242e49ebe
How to cite
Covariance-aware Scoring: A Geometric Correction for Vision-Language Test-Time Adaptation — arXiv:2611.00024 — arXiv (en), https://arxiv.org/abs/2611.00024 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_9cd242e49ebe (Metraipolis archive dc_9cd242e49ebe, sha256 4d415dd536031052d70f658b2521fac67c7b553322e7a24ddd2d191dcce537a8)
BibTeX
@misc{dc9cd242e49ebe,
title = {Covariance-aware Scoring: A Geometric Correction for Vision-Language Test-Time Adaptation — arXiv:2611.00024},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.00024},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_9cd242e49ebe; sha256 4d415dd536031052d70f658b2521fac67c7b553322e7a24ddd2d191dcce537a8},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.00024. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Test-time adaptation (TTA) enables vision--language models such as CLIP to adapt to distribution shifts without labels or access to source data. Existing methods adapt through feature caching, prototype estimation, or the online accumulation of feature statistics, but most ultimately score a test feature against a class representative in the ambient embedding space, typically through cosine similarity. We identify this shared decision rule as a limitation under anisotropic feature distributions, which means when a small number of directions account for a large fraction of feature variance, the score can be dominated by high-variance directions that are not necessarily discriminative. In light of this, we propose CARe, a covariance-aware scoring for TTA that replaces the ambient geometry with a Mahalanobis geometry induced by the target feature second moment. Theoretically, we show that under a shared-covariance Gaussian model the covariance-aware decision direction coincides with that of the Bayes-optimal linear discriminant, with ambient-space scoring as the isotropic special case, and establish that the potential benefit of covariance-aware scoring grows with feature anisotropy. Empirically, we show that CARe improves adaptation accuracy over its base cosine scoring on both standard and fine-grained distribution-shift benchmarks across two backbones. Consistent with our analysis, the gains are largest on datasets whose features are most anisotropic and vanish as the features approach isotropy. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2611.00024 — 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 2026-11-03T11:02:37.740Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).