Personalized Decentralized Federated Learning with Knowledge Distillation

被引:8
作者
Jeong, Eunjeong [1 ]
Kountouris, Marios [1 ]
机构
[1] EURECOM, Commun Syst Dept, F-06410 Sophia Antipolis, France
来源
ICC 2023-IEEE INTERNATIONAL CONFERENCE ON COMMUNICATIONS | 2023年
关键词
decentralized federated learning; personalization; knowledge distillation;
D O I
10.1109/ICC45041.2023.10279714
中图分类号
TN [电子技术、通信技术];
学科分类号
0809 ;
摘要
Personalization in federated learning (FL) functions as a coordinator for clients with high variance in data or behavior. Ensuring the convergence of these clients' models relies on how closely users collaborate with those with similar patterns or preferences. However, it is generally challenging to quantify similarity under limited knowledge about other users' models given to users in a decentralized network. To cope with this issue, we propose a personalized and fully decentralized FL algorithm, leveraging knowledge distillation techniques to empower each device so as to discern statistical distances between local models. Each client device can enhance its performance without sharing local data by estimating the similarity between two intermediate outputs from feeding local samples as in knowledge distillation. Our empirical studies demonstrate that the proposed algorithm improves the test accuracy of clients in fewer iterations under highly non-independent and identically distributed (non-i.i.d.) data distributions and is beneficial to agents with small datasets, even without the need for a central server.
引用
收藏
页码:1982 / 1987
页数:6
相关论文
共 29 条
[1]  
Anil R., 2018, INT C LEARN REPR, DOI DOI 10.1109/TVLSI.2020.2981443
[2]  
Bellet A, 2018, PR MACH LEARN RES, V84
[3]  
Borodich E., 2021, WORKSH NEW FRONT FED
[4]  
Divi S., 2021, ARXIV210515191
[5]  
Fallah Alireza, 2020, Personalized federated learning: A meta-learning approach
[6]  
Geiping Jonas, 2020, ADV NEURAL INFORM PR, V33
[7]  
Hinton G., 2015, ARXIV
[8]  
Huang Y., 2020, PERSONALIZED FEDERAT
[9]   A Survey on Federated Learning for Resource-Constrained IoT Devices [J].
Imteaj, Ahmed ;
Thakker, Urmish ;
Wang, Shiqiang ;
Li, Jian ;
Amini, M. Hadi .
IEEE INTERNET OF THINGS JOURNAL, 2022, 9 (01) :1-24
[10]  
Jeong E., 2018, WORKSH MACH LEARN PH