Accelerating Distributed Deep Learning By Adaptive Gradient Quantization | IEEE Conference Publication | IEEE Xplore