New Efficient Pruning Algorithms For Compressing And Accelerating Convolutional Neural Networks.