Reliability-Gated Source Anchoring for Continual Test-Time Adaptation

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Singh, Vikash, Ganguly, Debargha, Chen, Weicong, Sahoo, Sabyasachi, Sankar, Sreehari, Zhang, Biyao, Hariri, Mohsen, Wang, Shouren, Zafar, Osama, Gagné, Christian, Chaudhary, Vipin
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866910233976111104
author Singh, Vikash
Ganguly, Debargha
Chen, Weicong
Sahoo, Sabyasachi
Sankar, Sreehari
Zhang, Biyao
Hariri, Mohsen
Wang, Shouren
Zafar, Osama
Gagné, Christian
Chaudhary, Vipin
author_facet Singh, Vikash
Ganguly, Debargha
Chen, Weicong
Sahoo, Sabyasachi
Sankar, Sreehari
Zhang, Biyao
Hariri, Mohsen
Wang, Shouren
Zafar, Osama
Gagné, Christian
Chaudhary, Vipin
contents Continual test-time adaptation (CTTA) updates a pretrained model online on an unlabeled, non-stationary stream while anchoring it to a frozen source checkpoint. This anchor is useful only when the source remains reliable. On CCC-Hard, however, a ResNet-50 source falls to approximately $1.3\%$ top-$1$ accuracy, while existing source-anchored CTTA methods continue applying the same anchor strength. We call this failure mode blind anchoring and propose RMemSafe, a reliability-gated extension of ROID that uses the frozen source's normalized predictive entropy to attenuate all explicit source-coupled uses in the objective. When the source posterior approaches uniformity, the gate closes: the source anchor and agreement filter vanish, and the objective reduces to a source-agnostic fallback comprising ROID's base losses plus marginal calibration. Combined with ASR, RMemSafe achieves the lowest error on $8$ of $9$ matched-split continual-corruption cells and is the best reset-based method on all $9$, improving ROID+ASR by $1.05$~pp on ResNet-50 and $0.48$~pp on ViT-B/16. A controlled source-degradation sweep shows a $1.13{\times}$ shallower harm slope than ROID+ASR, consistent with the graceful-decay prediction. The entropy gate detects high-entropy source collapse, not confidently wrong low-entropy sources; this scope is explicitly evaluated and discussed.
format Preprint
id arxiv_https___arxiv_org_abs_2605_14063
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Reliability-Gated Source Anchoring for Continual Test-Time Adaptation
Singh, Vikash
Ganguly, Debargha
Chen, Weicong
Sahoo, Sabyasachi
Sankar, Sreehari
Zhang, Biyao
Hariri, Mohsen
Wang, Shouren
Zafar, Osama
Gagné, Christian
Chaudhary, Vipin
Machine Learning
Continual test-time adaptation (CTTA) updates a pretrained model online on an unlabeled, non-stationary stream while anchoring it to a frozen source checkpoint. This anchor is useful only when the source remains reliable. On CCC-Hard, however, a ResNet-50 source falls to approximately $1.3\%$ top-$1$ accuracy, while existing source-anchored CTTA methods continue applying the same anchor strength. We call this failure mode blind anchoring and propose RMemSafe, a reliability-gated extension of ROID that uses the frozen source's normalized predictive entropy to attenuate all explicit source-coupled uses in the objective. When the source posterior approaches uniformity, the gate closes: the source anchor and agreement filter vanish, and the objective reduces to a source-agnostic fallback comprising ROID's base losses plus marginal calibration. Combined with ASR, RMemSafe achieves the lowest error on $8$ of $9$ matched-split continual-corruption cells and is the best reset-based method on all $9$, improving ROID+ASR by $1.05$~pp on ResNet-50 and $0.48$~pp on ViT-B/16. A controlled source-degradation sweep shows a $1.13{\times}$ shallower harm slope than ROID+ASR, consistent with the graceful-decay prediction. The entropy gate detects high-entropy source collapse, not confidently wrong low-entropy sources; this scope is explicitly evaluated and discussed.
title Reliability-Gated Source Anchoring for Continual Test-Time Adaptation
topic Machine Learning
url https://arxiv.org/abs/2605.14063