Variance-based Gradient Compression for Efficient Distributed Deep Learning.

Yusuke Tsuzuku, Hiroto Imachi, Takuya Akiba

Browse the full ICLR paper archive.