1-bit stochastic gradient descent and its application to data-parallel distributed training of speech DNNs.

Frank Seide, Hao Fu, Jasha Droppo, Gang Li, Dong Yu

Browse the full Interspeech paper archive.