Hybrid Convolution Architecture for Energy-Efficient Deep Neural Network Processing | IEEE Journals & Magazine | IEEE Xplore