让医学深度学习训练结果完全可复现，消除随机性干扰。
Bit-Identical Medical Deep Learning via Structured Orthogonal Initialization
- 用结构正交初始化、黄金比例排序和自定义自动微分消除三类随机源。
- 在罕见心律分类上，个体预测差异降低7.5倍，AUC波动从30.9降至4.1个百分点。
- 复现性经MD5验证，且不损失常规任务性能，适合医疗模型可信验证。
深度学习训练具有非确定性：相同代码使用不同随机种子会得到聚合指标一致但个体预测不同的模型，尤其在罕见临床类别上，每类AUC波动超过20个百分点。本文提出一种可验证的比特级完全复现训练框架，通过结构正交基函数初始化、黄金比例批序调度以及架构选择与定制autograd，消除权重初始化、批次顺序和非确定性GPU运算三类随机性。该流程可生成跨独立运行的MD5校验一致的训练权重。在PTB-XL心电图节律分类任务中，结构化初始化显著优于Kaiming（n=20；Conformer p=0.016，Baseline p<0.001），使整体方差降低2-3倍，罕见节律的个体预测变异性最高减少7.5倍（TRIGU范围：4.1pp vs 30.9pp，经三重交叉验证确认）。四基函数对比（n=20）显示所有结构正交基表现等效（Friedman p=0.48），证明贡献来自结构化初始化本身，而非特定基函数。七项MedMNIST跨域验证（n=20，全部p>0.14）表明标准任务无性能损失；对不平衡任务（ChestMNIST、RetinaMNIST）的类别分析显示，罕见类别方差同样显著降低。在三个外部心电图数据库上的零样本泛化测试中，房颤分类AUC均超过0.93。
原文摘要 · Abstract (English)
Deep learning training is non-deterministic: identical code with different random seeds produces models that agree on aggregate metrics but disagree on individual predictions, with per-class AUC swings exceeding 20 percentage points on rare clinical classes. We present a framework for verified bit-identical training that eliminates three sources of randomness: weight initialization (via structured orthogonal basis functions), batch ordering (via golden ratio scheduling), and non-deterministic GPU operations (via architecture selection and custom autograd). The pipeline produces MD5-verified identical trained weights across independent runs. On PTB-XL ECG rhythm classification, structured initialization significantly exceeds Kaiming across two architectures (n=20; Conformer p = 0.016, Baseline p < 0.001), reducing aggregate variance by 2-3x and reducing per-class variability on rare rhythms by up to 7.5x (TRIGU range: 4.1pp vs 30.9pp under Kaiming, independently confirmed by 3-fold CV). A four-basis comparison at n=20 shows all structured orthogonal bases produce equivalent performance (Friedman p=0.48), establishing that the contribution is deterministic structured initialization itself, not any particular basis function. Cross-domain validation on seven MedMNIST benchmarks (n=20, all p > 0.14) confirms no performance penalty on standard tasks; per-class analysis on imbalanced tasks (ChestMNIST, RetinaMNIST) shows the same variance reduction on rare classes observed in ECG. Cross-dataset evaluation on three external ECG databases confirms zero-shot generalization (>0.93 AFIB AUC).
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。