arXiv:2605.14063cs.LG2026-05被引 1

提出可靠性门控锚点机制,让模型在源失效时自动放弃依赖,避免错误累积。

Reliability-Gated Source Anchoring for Continual Test-Time Adaptation

  • 用源模型预测熵判断其可靠性,动态关闭锚点强度
  • 在9个持续退化场景中8个表现最优,比基线降低1.05个百分点误差
  • 适合源数据可能失效的持续学习场景,尤其关注鲁棒性

持续测试时自适应(CTTA)在无标签、非平稳的数据流上在线更新预训练模型,同时将其锚定于一个冻结的源检查点。但当源模型可靠性下降时,现有方法仍维持固定锚点强度,导致性能恶化。本文将此现象称为‘盲锚’,提出RMemSafe:基于ROID的可靠性门控扩展,利用冻结源模型的归一化预测熵来抑制所有显式源耦合项。当源后验趋于均匀时,门控关闭——源锚点与一致性过滤器消失,目标函数退化为不含源依赖的备选方案,包含ROID基础损失与边缘校准项。结合ASR,RMemSafe在9个匹配分割的持续退化单元中取得8个最低误差,是所有9个单元中最佳的重置型方法,在ResNet-50上比ROID+ASR降低1.05个百分点,在ViT-B/16上降低0.48个百分点。受控源退化实验显示其误差恶化斜率比ROID+ASR低1.13倍,符合渐进衰减预期。熵门控能识别高熵源崩溃,而非误判低熵但错误的源;该检测范围明确评估并讨论。

原文摘要 · Abstract (English)

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.

持续学习测试时适应可靠性评估模型鲁棒性

Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品;中文卡片由大模型生成,请以原文为准。