卷积神经网络模型压缩技术(Hardwareoriented Approximation of Convolutional Neural Networks)
HighcomputationalcomplexityhindersthewidespreadusageofConvolutionalNeuralNetworks(CNNs),especiallyinmobiledevices.Hardwareacceleratorsarearguablythemostpromisingapproachforreducingbothexecutiontimeandpowerconsumption.Oneofthemostimportantstepsinacceleratordeve