QCS-CNN: A Neural Network Optimization Algorithm in Resource-Constrained Environment
Di Liu, Zhizhao Feng · 2020
Due to the complex models and large amount of calculation, deep neural networks cannot be deployed on resource-constrained devices. In order to apply deep neural networks to resource-constrained devices, this paper proposes a QCS-CNN neural network optimization algorithm. It can compress the input image, while reducing the number of parameters, it can quantize the precision of all parameters in the network from 32 bits to 8 bits. The experiments on MNIST data set show that QCS-CNN neural network optimization algorithm can effectively reduce the amount of calculation and accelerate the inference speed of the model.