arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1812.04048cs.DCcs.SYeess.SY
收藏

压缩分布式梯度下降:通信高效的网络共识

Compressed Distributed Gradient Descent: Communication-Efficient Consensus over Networks

Iowa State University(爱荷华州立大学);Air Force Research Laboratory, Information Directorate(空军研究实验室信息管理局)

Xin Zhang, Jia Liu, Zhengyuan Zhu, Elizabeth S. Bentley

AI总结:

本文提出了一种通信高效的DGD类型算法ADC-DGD,解决了高维网络共识问题中的通信瓶颈问题,并证明了其收敛性与无压缩DGD相同,揭示了收敛速度的相变现象。

摘要:

近年来,网络共识优化问题受到越来越多的关注,并在许多科学和工程领域中找到了重要的应用。为了解决网络共识优化问题,最著名的做法之一是分布式梯度下降方法(DGD)。然而,在通信速率较慢的网络中,DGD在解决高维网络共识问题时表现不佳,由于存在通信瓶颈。这促使我们设计了一种基于压缩信息交换的通信高效的DGD类型算法。本文的贡献有三方面:i)我们开发了一种称为放大微分压缩DGD(ADC-DGD)的通信高效算法,并证明其在任何无偏压缩算子下都能收敛;ii)我们严格证明了ADC-DGD的收敛性能,并证明其与无压缩DGD的性能相同;iii)我们揭示了ADC-DGD收敛速度中的有趣相变现象。总体而言,我们的发现推动了网络共识优化理论的发展。

英文摘要:

Network consensus optimization has received increasing attention in recent years and has found important applications in many scientific and engineering fields. To solve network consensus optimization problems, one of the most well-known approaches is the distributed gradient descent method (DGD). However, in networks with slow communication rates, DGD's performance is unsatisfactory for solving high-dimensional network consensus problems due to the communication bottleneck. This motivates us to design a communication-efficient DGD-type algorithm based on compressed information exchanges. Our contributions in this paper are three-fold: i) We develop a communication-efficient algorithm called amplified-differential compression DGD (ADC-DGD) and show that it converges under {\em any} unbiased compression operator; ii) We rigorously prove the convergence performances of ADC-DGD and show that they match with those of DGD without compression; iii) We reveal an interesting phase transition phenomenon in the convergence speed of ADC-DGD. Collectively, our findings advance the state-of-the-art of network consensus optimization theory.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑