SuperNeurons: FFT-based Gradient Sparsification in the Distributed Training of Deep Neural Networks

GNN
Re-assign community