Distributed SignSGD With Improved Accuracy and Network-Fault Tolerance

Distributed SignSGD With Improved Accuracy and Network-Fault Tolerance

This paper proposes DropSignSGD, a communication-efficient and network-fault tolerant algorithm for training deep neural networks in a distributed and synchronous fashion. In DropSignSGD, all numerical elements communicated between machines are either 1 or -1, represented by only one bit. More impor...

Full description

Bibliographic Details
Main Authors: Trieu Le Phong, Tran Thi Phuong
Format: Article
Language:English
Published: IEEE 2020-01-01
Series:IEEE Access
Subjects:
Network-fault tolerance
communication efficiency
distributed SGD
deep learning
Online Access:https://ieeexplore.ieee.org/document/9234512/

Internet

https://ieeexplore.ieee.org/document/9234512/

Similar Items