An Energy-Efficient Deep Convolutional Neural Network Inference Processor With Enhanced Output Stationary Dataflow in 65-nm CMOS | IEEE Journals & Magazine | IEEE Xplore