面向非独立同分布数据的联邦学习研究进展

Advances in Federated Learning for Non-independent Identically Distributed Data

在联邦学习中,因数据只需要在终端设备上进行训练而不需要上传到服务器端,从而有效的保证了数据的隐私性和安全性.目前关于联邦学习各方面的研究虽然取得了很大的进展,但是联邦学习中的非独立同分布数据问题仍然是一个难以攻克的难题.本文对联邦学习中非独立同分布数据方面的研究进行了大量的调研,发现现有的研究主要涉及以下几个方面:性能优化、算法优化、模型优化、通信成本、隐私保护和个性化联邦学习等.为了归纳整理联邦学习中关于非独立同分布数据的相关研究,本文从以上各个方面详细介绍了现阶段联邦学习中有关非独立同分布数据的研究方案;最后分析了联邦学习中非独立同分布数据未来的研究方向,为今后联邦学习的研究指明方向.

In federated learning,data only needs to be trained on the terminal device without being uploaded to the server,thus effectively ensuring the privacy and security of data.Although great progress has been made in all aspects of federated learning,the problem of non-Independent Identically Distributed data in federated learning is still a difficult problem to overcome.In this paper,a lot of research on non-Independent Identically Distributed data in federated learning is conducted,and it is found that the existing research mainly involves the following aspects:performance optimization,algorithm optimization,model optimization,communication cost,privacy protection and personalized federated learning.In order to summarize the relevant researches on non-Independent Identically Distributed data in federal learning,this paper introduces in detail the current research programs on non-Independent Identically Distributed data in federal learning from the above aspects.Finally,the future research direction of non-Independent Identically Distributed data in federal learning is analyzed to point out the future research direction of federal learning.