提出PSDNorm，提升睡眠分期模型在新数据上的泛化能力
PSDNorm: Test-Time Temporal Normalization for Deep Learning in Sleep Staging
- 利用时序上下文和Monge映射进行特征归一化
- 在10个数据集上表现超越现有方法，尤其在数据稀缺时更稳定
- 适合需要跨机构泛化的睡眠信号分析场景
分布偏移是机器学习中的重大挑战，尤其在跨受试者、机构和设备采集的生物医学数据（如睡眠数据）中。现有的归一化层（如BatchNorm、LayerNorm、InstanceNorm）在时间维度应用时，忽略了特征向量内在的依赖关系与自相关性。本文提出PSDNorm，通过利用Monge映射和时序上下文来对深度学习模型中的信号特征图进行归一化。在基于U-Net或Transformer骨干网络、训练于10,000名受试者及10个数据集的模型上评估表明，PSDNorm在未见数据集上达到当前最优性能，且在数据稀缺条件下更具鲁棒性。
原文摘要 · Abstract (English)
Distribution shift poses a significant challenge in machine learning, particularly in biomedical applications using data collected across different subjects, institutions, and recording devices, such as sleep data. While existing normalization layers, BatchNorm, LayerNorm and InstanceNorm, help mitigate distribution shifts, when applied over the time dimension they ignore the dependencies and auto-correlation inherent to the vector coefficients they normalize. In this paper, we propose PSDNorm that leverages Monge mapping and temporal context to normalize feature maps in deep learning models for signals. Evaluations with architectures based on U-Net or transformer backbones trained on 10K subjects across 10 datasets, show that PSDNorm achieves state-of-the-art performance on unseen left-out datasets while being more robust to data scarcity.
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。