Metraipolis › archive › dc_528d59902dc2

Adapting Without Gradients: Affine Statistics Transport and What Its Certificate Can Tell You — arXiv:2609.00374

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.00374
License (as archived): CC0-1.0
Archived: 2026-10-17T06:33:18.129Z
SHA-256 of the archived text: 7cfc96dc80a14dcbefadd91bcc95b1f597c3bd45bb9fa3c522ec25395de08949
Stable URL of this page: https://metraipolis.com/documento/dc_528d59902dc2

How to cite

Adapting Without Gradients: Affine Statistics Transport and What Its Certificate Can Tell You — arXiv:2609.00374 — arXiv (en), https://arxiv.org/abs/2609.00374 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_528d59902dc2 (Metraipolis archive dc_528d59902dc2, sha256 7cfc96dc80a14dcbefadd91bcc95b1f597c3bd45bb9fa3c522ec25395de08949)

BibTeX

@misc{dc528d59902dc2,
  title = {Adapting Without Gradients: Affine Statistics Transport and What Its Certificate Can Tell You — arXiv:2609.00374},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.00374},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_528d59902dc2; sha256 7cfc96dc80a14dcbefadd91bcc95b1f597c3bd45bb9fa3c522ec25395de08949},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.00374. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Test-time adaptation (TTA) typically assumes that model parameters can be updated at inference time. This assumption is restrictive for inference-only accelerators, frozen or third-party models, and memory-constrained deployments, and standard BatchNorm-based TTA configurations may also become inactive on architectures without BatchNorm. We study adaptation when the learned model must remain frozen. We introduce CASTER, a gradient-free method that stores source class statistics in a discriminative subspace, estimates a class-shared affine transformation from target-batch moments, and analytically transports the source class distributions before classification. CASTER requires no backward pass, optimizer state, or stored source feature bank. Across four backbones and seven datasets, it outperforms k-NN on identical frozen features in 27 of 28 backbone-dataset settings while retaining a median of 18x less state. Affine transport is not always reliable. On ImageNet-C, where batches contain only 64 samples for 1000 classes, unconditional transport loses 21.2 top-1 points. We therefore introduce an empirical residual-to-margin transportability certificate. Across 307 evaluation cells, every transport losing more than 10 points has certificate value above 3.9, although benign and destructive regimes are not perfectly separated. Gating converts an average $-3.35$-point effect of unconditional transport into a +1.69-point gain, and performance remains within 0.3 points of the best threshold over a broad threshold range. Finally, we show that this certificate is mechanism-specific: when applied to Tent, it accepts only $4.3\%$ of updates and preserves 0.6% of Tent's available gain. These results position CASTER as a lightweight adaptation mechanism for frozen-model deployment, together with an explicit account of when its safety signal is informative and when it is not.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-02 in cs.AI).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.00374 — 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 2026-10-17T06:15:12.775Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-02?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).