Flattened one-bit stochastic gradient descent: compressed distributed optimization with controlled variance

Open in new window