On-Chip Memory Optimization of High Efficiency Accelerator for Deep Convolutional Neural Networks | IEEE Conference Publication | IEEE Xplore