Sorry, you need to enable JavaScript to visit this website.

Q-GADMM: Quantized Group ADMM for Communication Efficient Decentralized Machine Learning

Citation Author(s):
Anis Elgabli, Jihong Park, Amrit Bedi, Mehdi Bennis, Vaneet Aggarwal
Submitted by:
Anis Elgabli
Last updated:
21 May 2020 - 3:34pm
Document Type:
Presentation Slides
Document Year:
Event:
Presenters Name:
Anis A Elgabli

Abstract 

Abstract: 

In this paper, we propose a communication-efficient decentralized machine learning (ML) algorithm, coined quantized group ADMM (Q-GADMM). Every worker in Q-GADMM communicates only with two neighbors, and updates its model via the group alternating direct method of multiplier (GADMM), thereby ensuring fast convergence while reducing the number of communication rounds. Furthermore, each worker quantizes its model updates before transmissions, thereby decreasing the communication payload sizes. We prove that Q-GADMM converges to the optimal solution for convex loss functions, and numerically show that Q-GADMM yields 7x less communication cost while achieving almost the same accuracy and convergence speed compared to GADMM without quantization.

up
0 users have voted:

Dataset Files

icassp2020_final.pdf

(118)