动态追踪训练中样本影响变化，发现关键样本随阶段更替而改变作用。
Dynamic Influence Tracker: Measuring Time-Varying Sample Influence During Training
- 基于时间窗口分析样本影响演化，揭示学习阶段转移规律。
- 在复杂架构中检测污染样本准确率达98%以上，相关性达0.99。
- 无需假设损失凸性或模型收敛，适用于多种训练场景。
现有方法仅提供静态的训练样本影响度量，忽略了样本影响随训练过程的变化。本文提出动态影响追踪器（Dynamic Influence Tracker, DIT），可捕捉训练过程中任意时间窗口的时变样本影响。DIT揭示三个关键现象：1）样本具有不同的时变影响模式，部分样本在训练初期重要，而另一些则在后期才显现影响；2）早期与晚期的影响呈弱相关，表明模型经历不同学习阶段且关注重点不断变化；3）在收敛期分析影响，比全周期分析更高效、准确地检测出污染样本。该方法具备理论保证，不依赖损失凸性或模型收敛假设，显著优于现有方法，在复杂架构中达到高达0.99的相关性与超过98%的污染样本检测准确率。
原文摘要 · Abstract (English)
Existing methods for measuring training sample influence on models only provide static, overall measurements, overlooking how sample influence changes during training. We propose Dynamic Influence Tracker (DIT), which captures the time-varying sample influence across arbitrary time windows during training. DIT offers three key insights: 1) Samples show different time-varying influence patterns, with some samples important in the early training stage while others become important later. 2) Sample influences show a weak correlation between early and late stages, demonstrating that the model undergoes distinct learning phases with shifting priorities. 3) Analyzing influence during the convergence period provides more efficient and accurate detection of corrupted samples than full-training analysis. Supported by theoretical guarantees without assuming loss convexity or model convergence, DIT significantly outperforms existing methods, achieving up to 0.99 correlation with ground truth and above 98\% accuracy in detecting corrupted samples in complex architectures.
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。