DOI number:10.1109/INFOCOMWKSHPS61880.2024.10620731
Affiliation of Author(s):杭州国际创新研究院
Journal:IEEE INFOCOM 2024 - IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS)
Place of Publication:加拿大温哥华
Key Words:hierarchical federated learning; loss-based heterogeneity; aggregation weights; mobile edge computing; non-IID data
Abstract:分析无线层次化联邦学习在MEC内独立同分布、MEC间非独立同分布场景下的数据异质性和全局聚合间隔影响，提出基于损失异质性的聚合权重设计，以加快训练并提升精度；在高异质性场景中增益更为明显。
Note:IEEE INFOCOM 2024 Workshops论文；会议地点为加拿大温哥华。
Co-author:Youjia Chen, Junnan Yang, Ming Ding, Peng Cheng, Jinsong Hu, Haifeng Zheng
First Author:Yuchuan Ye
Indexed by:会议论文
Document Code:10620731
Discipline:工学
First-Level Discipline:Information and Communication Engineering
Document Type:EI会议
Page Number:1-6
Translation or Not:no
Date of Publication:2024-05-20