发表机构
KTH Royal Institute of Technology; Shanghai Jiao Tong University; Global Institute of Future Technology（瑞典皇家理工学院; 上海交通大学; 未来技术全球研究院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究剖析时间表格漂移下冻结基础模型序贯适应的收益来源与极限，识别四种机制并界定不可约的识别墙，提出估计方法，实验验证有限样本表现。
AI 中文摘要
在时间漂移下，对冻结的表格基础模型进行序贯适应（每个标签仅在预测后揭示）有助于某些部署，却损害其他部署，而当前实践无法预测哪种情况会发生。我们研究了这些收益的来源与极限。一种诊断性解剖将收益归因于流式协议下的四种反复出现的机制，该协议消除了三种乐观偏差并量化了第四种。在不可知的总体变差漂移类别中，目标条件分布仅被部分识别：其识别集直径（即“墙”）在样本量上均匀地无法从无标签数据中缩减。第二个正交的 $L^2$ 投影墙量化了冻结表示无法表达的内容。两种典型机制先验坍缩了第一堵墙。在给定的干扰率条件下，该墙可以从带标签的历史窗口中以高于边际阈值 $\u03b3^\u2217=d_0/(2\u03b1_s)$ 的 $\u221a N$ 速率估计。在 $\u03b3=0$ 时，条件下界程序依赖于一个开放的亲和性估计；正边际下分支也仍然开放。半合成数据展示了具有校准指数的有限样本机制。在给定的粗糙度界限下，八个工业流上的流级代理落在困难一侧，而等式情况 $\u03b3=\u03b3^\u2217$ 仍未解决。
英文摘要
Prequential adaptation of frozen tabular foundation models under temporal drift, with each label revealed only after prediction, helps some deployments and harms others, yet current practice does not predict which. We study the sources and limits of these gains. A diagnostic anatomy attributes gains to four recurring mechanisms under a streaming protocol that removes three optimistic biases and quantifies a fourth. Within an agnostic total-variation drift class, the target conditional is only partially identified: its identified-set diameter, the \emph{wall}, is irreducible from unlabeled data uniformly in sample size. A second, orthogonal $L^2$ projection wall quantifies what the frozen representation cannot express. Two canonical mechanism priors collapse the first wall. Under stated nuisance-rate conditions, the wall can be estimated from labeled historical windows at a $\sqrt N$ rate above the margin threshold $γ^\star=d_0/(2α_s)$. At $γ=0$, the conditional lower-bound program depends on an open affinity estimate; the positive-margin lower branch also remains open. Semi-synthetic data illustrate the finite-sample mechanism with calibrated exponents. Stream-level proxies on eight industrial streams fall on the difficult side under a stated roughness bound, while the equality case $γ=γ^\star$ remains unresolved.
