ResNet微调中类别分离跃迁次数可预测测试准确率，越少越准。
Phase Transition Frequency as a Training Time Predictor of Test Accuracy in ResNets
- 统计微调过程中的类别分离跃迁次数作为预测指标。
- 在CIFAR-10/100上相关系数达-0.84和-0.87，显著负相关。
- 适合用于训练日志中低成本监控模型质量，尤其在分布内任务。
本文通过75组实验（涵盖CIFAR-10、CIFAR-100、TinyImageNet和CIFAR-10-C四个基准，ResNet-18/50/101三种架构，每配置5~10个随机种子），实证分析了ResNet微调过程中观察到的离散类别可分性跃迁次数对最终测试准确率的预测能力。在标准i.i.d.分类任务中，其与测试精度呈强负相关：CIFAR-10上r = -0.84（p < 10⁻⁸, n = 30），CIFAR-100上r = -0.87（p < 10⁻⁵, n = 15）。在分布外压力下，相关性减弱：TinyImageNet上r = -0.45，CIFAR-10-C上r = -0.19。部分相关性分析显示，控制网络深度后，其在CIFAR-100上仍具统计显著性（r_partial = -0.69, p = 0.007）。与其他六种训练曲线信号对比，跃迁次数在CIFAR-100上相关性最强，在CIFAR-10上为前列，但在受压数据集上被其他信号超越。该方法可作为分布内训练质量的廉价探针，适用于标准训练流程中实时日志记录。
原文摘要 · Abstract (English)
The number of discrete class-separability jumps observed during ResNet finetuning is examined empirically as a predictor of final test accuracy. Across 75 experiments spanning four benchmarks (CIFAR-10, CIFAR-100, TinyImageNet, and CIFAR-10-C) and three architectures (ResNet-18, ResNet-50, and ResNet-101), with five to ten seeds per configuration, a strong within-dataset negative correlation is obtained on standard i.i.d. classification benchmarks: \(r = -0.84\) on CIFAR-10 (\(p < 10^{-8}\), \(n = 30\)) and \(r = -0.87\) on CIFAR-100 (\(p < 10^{-5}\), \(n = 15\)). Under distributional stress, the relationship attenuates: TinyImageNet yields \(r = -0.45\), and the CIFAR-10-C corruption benchmark yields \(r = -0.19\). Two additional analyses discipline the empirical claim. A partial correlation controlling for architecture depth, treated as a linear covariate, shows that on CIFAR-100 the transition count retains statistically significant predictive power (\(r_{\mathrm{partial}} = -0.69\), \(p = 0.007\)); the corresponding result under the stricter categorical conditioning is not established at \(n = 15\). A comparison against six alternative training-curve signals shows that transition count achieved the strongest correlation among the evaluated signals on CIFAR-100 and one of the strongest on CIFAR-10, but is dominated by other signals on the two stressed benchmarks. The comparison is restricted to training-curve-level signals; comparisons against effective rank, Hessian sharpness, Fisher information, margin, and neural-collapse measures, which are the strongest competitors in the current literature, are not part of the present study and remain open. The observation is presented as an in-distribution training-quality probe among a family of candidate probes, and an inexpensive detection procedure suitable for logging alongside a standard training loop is provided.
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。