gradient descent operation

EF21-P and friends: Improved theoretical communication complexity for distributed optimization with bidirectional compression

Feb 6, 12:00 - 13:00

B9 L2 H2 H2

EF21-P distributed optimization gradient descent operation

In this work we focus our attention on distributed optimization problems in the context where the communication time between the server and the workers is non-negligible. We obtain novel methods supporting bidirectional compression (both from the server to the workers and vice versa) that enjoy new state-of-the-art theoretical communication complexity for convex and nonconvex problems.