[论文] Prospective Prediction of OOD Degradation from Source-Side Training Dy...

论文概要

研究领域: ML
作者: Sasha, Monin
发布时间: 2026-10-08
arXiv: 2610.12397

中文摘要

我们研究能否在直接观测到分布外（OOD）性能退化之前，仅利用源端训练动态对其进行预测。在一个受控的捷径学习场景中，一个简单的逻辑回归预测器能够发展出明确的前瞻性信号，而仅使用训练时间则做不到。源端量的时序汇总统计比其当前值包含更多信息。当不做任何额外训练、将预测器从 CNN 迁移到 MLP 时，置信度和熵的动态仍保留了大量预测信息。这些结果从原理上证明：源端训练动态中蕴含着 OOD 失效的早期预警信号。

原文摘要

We study whether persistent out-of-distribution (OOD) degradation can be predicted before it is directly observed using only source-side training dynamics. In a controlled shortcut-learning setting, a simple logistic regression predictor develops a clear prospective signal, while training time alone does not. Temporal summaries of the source-side quantities are substantially more informative than their current values. When transferred without additional training from a CNN to an MLP, confidence and entropy dynamics retain substantial predictive information. These results provide a proof of principle that source-side training dynamics can contain an early warning signal for future OOD failure.

自动采集于 2026-10-12

#论文 #arXiv #ML #小凯