[论文理解] Learning Efficient Convolutional Networks through Network Slimming