摘要
Federated learning aims to collaboratively train a machine learning model with possibly geo-distributed workers,which is inherently communication constrained.To achieve communication efficiency,the conventional federated learning algorithms allow the worker to decrease the communication frequency by training the model locally for multiple times.Conventional federated learning architecture,inherited from the parameter server design,relies on highly centralised topologies and large nodes-to-server bandwidths,and convergence property relies on the stochastic gradient descent training in local,which usually causes the large end-to-end training latency in real-world federated learning scenarios.Thus,in this study,the authors propose the adaptive partial gradient aggregation method,a gradient partial level decentralised federated learning,to tackle this problem.In FedPGA,they propose a partial gradient exchange mechanism that makes full use of node-to-node bandwidth for speeding up the communication time.Besides,an adaptive model updating method further reduces the convergence rate by adaptive increasing the step size of the stable direction of gradient descent.The experimental results on various datasets demonstrate that the training time is reduced up to 14×compared to baselines without accuracy degrade.
基金
This work is funded by:National Key R&D Plan of China under Grant No.2017YFA0604500
by National Sci-Tech Support Plan of China under Grant No.2014BAH02F00
by National Natural Science Foundation of China under Grant No.61701190
by Youth Science Foundation of Jilin Province of China under Grant No.20160520011JH&20180520021JH
by Youth Sci-Tech Innovation Leader and Team Project of Jilin Province of China under Grant No.20170519017JH
by Key Technology Innovation Cooperation Project of Government and University for the whole Industry Demonstration under Grant No.SXGJSF2017-4
by Key scientific and technological R&D Plan of Jilin Province of China under Grant No.20180201103GX
by Project of Jilin Province Development and Reform Commission under Grant No.2019FGWTZC001.
作者简介
Liang Hu,E-mail:hul@jlu.edu.cn。
