MARINA: Faster Non-Convex Distributed Learning with Compression.

Eduard Gorbunov, Konstantin Burlachenko, Zhize Li, Peter Richtrik

Browse the full ICML paper archive.