Reduce Computational Complexity for Convolutional Layers by Skipping Zeros

Open in new window