Fed-ECG

该数据集名为Fed-ECG，由七家机构提供的心血管疾病数据构建而成，针对联邦学习场景提出了重大挑战。由于数据呈现非独立同分布（non-IID）和长尾分布的特点，该数据集突显了在联邦学习过程中处理这两类数据分布的难题。该数据集适用于多中心联邦学习场景，其任务是对心电图（ECG）进行分类。The dataset is named Fed-ECG, which is constructed using cardiovascular disease data provided by seven institutions and poses significant challenges for federated learning scenarios. Due to the characteristics of non-independent and identically distributed (non-IID) and long-tailed distribution of the data, this dataset highlights the difficulties in handling these two types of data distributions during the federated learning process. This dataset is applicable for multi-center federated learning scenarios, with the task focusing on electrocardiogram (ECG) classification.

搜集汇总

背景与挑战

背景概述

Fed-ECG是首个真实世界联邦学习基准数据集，专注于心血管疾病数据，包含来自四个客户端（SPH、PTB-XL、SXPH、G12EC）的心电图数据。该数据集专为联邦学习场景设计，支持算法和模型的扩展，适用于多中心医疗数据协作研究。