文献类型：专利

英文题名：Convolutional neural network based image classification method, involves performing network parameters over-fitting phenomenon, optimizing momentum-related network parameters, and reducing network training difficulties

作者：WNG L;HE Y;WANG K;JIANG N;ZHONG G;CAI Y

机构：[1]UNIV EAST CHINA SCI & TECHNOLOGY

申请号：CN107341518-A

申请日：2017-07-07

公开日：2017-11-10

语种：英文

收录：DERWENT

摘要：NOVELTY - The method involves constructing a deep convolutional neural network. Function is activated according to a deep convolutional neural network. Deep convolutional neural network training and testing process is performed. Forward and backward propagating process is performed according to the deep convolutional neural network. Network parameters over-fitting phenomenon optimizing process is performed through a batch. Momentum-related network parameters are optimized. Network training difficulties are reduced. USE - Convolutional neural network based image classification method. ADVANTAGE - The method enables effectively improving recognition rate of the deep convolutional neural network, thus improving image classification accuracy. DESCRIPTION OF DRAWING(S) - The drawing shows a schematic flow diagram illustrating a convolutional neural network based image classification method. '(Drawing includes non-English language text)'