网络压缩(model compression) A Deep Neural Network Compression Pipeline: Pruning, Quantization, Huffman
论文《ADeepNeuralNetworkCompressionPipeline:Pruning,Quantization,HuffmanEncoding》Pruningbylearningonlytheimportantconnections.allconnectionswithweightsbelowathresholdareremovedfromthenetwork.retrainthene