Document worth reading: “A Survey on Acceleration of Deep Convolutional Neural Networks”

Deep Neural Networks have achieved excellent progress by way of the previous couple of years and are at current the fundamental devices of many intelligent strategies. At the similar time, the computational complexity and helpful useful resource consumption of these networks are moreover continuously rising. This will pose a serious drawback to the deployment of such networks, significantly for real-time functions or on resource-limited items. Thus, neighborhood acceleration have develop right into a scorching topic contained in the deep finding out group. As for {{hardware}} implementation of deep neural networks, a batch of accelerators based on FPGA/ASIC have been proposed these years. In this paper, we provide an entire survey regarding the present advances on neighborhood acceleration, compression and accelerator design from every algorithm and {{hardware}} side. Specifically, we provide thorough analysis for each of the following topics: neighborhood pruning, low-rank approximation, neighborhood quantization, teacher-student networks, compact neighborhood design and {{hardware}} accelerator. Finally, we make a dialogue and introduce only a few potential future directions. A Survey on Acceleration of Deep Convolutional Neural Networks