Title | ||
---|---|---|
CodedReduce: A Fast and Robust Framework for Gradient Aggregation in Distributed Learning |
Abstract | ||
---|---|---|
We focus on the commonly used synchronous Gradient Descent paradigm for large-scale distributed learning, for which there has been a growing interest to develop efficient and robust gradient aggregation strategies that overcome two key system bottlenecks: communication bandwidth and stragglers’ delays. In particular, Ring-AllReduce (RAR) design has been proposed to avoid bandwidth... |
Year | DOI | Venue |
---|---|---|
2019 | 10.1109/TNET.2021.3109097 | IEEE/ACM Transactions on Networking |
Keywords | DocType | Volume |
Bandwidth,Topology,Resilience,Distance learning,Computer aided instruction,Encoding,Training | Journal | 30 |
Issue | ISSN | Citations |
1 | 1063-6692 | 0 |
PageRank | References | Authors |
0.34 | 16 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Amirhossein Reisizadehmobarakeh | 1 | 61 | 6.44 |
Saurav Prakash | 2 | 1 | 0.69 |
Ramtin Pedarsani | 3 | 171 | 29.35 |
Amir Salman Avestimehr | 4 | 1880 | 157.39 |