Accelerating convolutional neural networks by group-wise 2D-filter pruning | IEEE Conference Publication | IEEE Xplore