[论文] When Should We Intervene at Test Time?

论文概要

研究领域: ML / 测试时适应 作者: Jiahao Zhang, Xinling Wen, Liangliang Li, Yu Chen 发布时间: 2026-10-27 arXiv: 2610.37658

中文摘要

测试时适应（TTA）帮助部署模型应对变化的输入，但同一种干预可能救一批数据、毁下一批。例如更新归一化统计量可以修复 corrupted 图像，但当批次中大部分属于同一类时就会失败。平均准确率掩盖了这些失败，也无法告诉模型何时该用干预。本文提出PORTAL——一个控制器，自主决定何时部署干预、给哪些输入打标签、何时停止查询。每个可用预测规则是一个候选者，其预测在当前反馈到达前就已固定，因此每个获取的标签都在相同输入上评估所有候选者。PORTAL结合这些共享比较，将查询导向仍有可能超越当前领先者的候选者，当剩余标签无法改变选择或潜在精度收益不再值得时停止。实验表明，PORTAL能将一个总体降低精度的干预转变为在选定批次上的有效工具，防止由样本到达顺序变化引起的严重失败。

原文摘要

Test-time adaptation helps a deployed model respond to changing inputs, but the same intervention can rescue one batch and damage the next. Updating normalization statistics, for example, can improve corrupted images yet fail when a batch contains mostly one class. Average accuracy hides these failures and cannot tell a model when to use an intervention. Target labels can guide this decision, but they are scarce and some choices are already clear. We introduce PORTAL, a controller that chooses when to deploy an intervention, which inputs to label, and when to stop querying. Each available prediction rule is a candidate, and its predictions are fixed before current feedback arrives. Every acquired label therefore evaluates all candidates on the same input. PORTAL combines these shared comparisons and directs queries toward candidates that can still overtake the current leader. It stops when the remaining labels cannot change the choice or when the possible accuracy gain no longer justifies another query. Our analysis links the required feedback to accuracy gaps and shared mistakes, explaining why some choices need much less evidence. Experiments show that this control turns an intervention that loses accuracy overall into a useful choice on selected batches. It prevents severe failures caused by changes in arrival order and preserves gains across complete corruption streams. When a fixed rule already works well, PORTAL uses little feedback to retain its performance. Periodic label checks reveal when past evidence needs refreshing and help the controller recover as conditions change. The acquired labels also train stronger candidates for future batches, and selection brings nearly all of their accuracy gains into deployment. The same feedback therefore improves both the choice made now and the predictions available to later batches.

---

*自动采集于 2026-10-29*

#论文 #arXiv #AI #小凯