Federated Learning Client Datasets
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/vdasu/deduplication
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了在联邦学习实验中使用的客户端数据集，旨在分析去重操作对模型性能的影响。数据集的大小可以从2的10次方变化到2的19次方，客户端数量可以从10个到50个不等，且设有30%的数据重复率。该数据集的规模具有多样性，涵盖了不同数据大小和客户端数量的情况，其研究任务专注于联邦学习。
This dataset comprises client datasets utilized in federated learning experiments, with the goal of analyzing the effect of deduplication operations on model performance. The dataset size ranges from 2^10 to 2^19, the number of clients varies between 10 and 50, and a 30% data duplication rate is configured. Boasting diverse scales, this dataset covers scenarios with different data sizes and client quantities, and its research task is specifically focused on federated learning.
提供机构：
Open-sourced by the authors
