Research on optimization method of convolutional nerual network
Xubin Feng, Xiuqin Su, Minqi Yan, Meilin Xie, Peng Liu, Xuezheng Lian, Feng Jing · 2018
With the improvement of computers' computation and storage performance, the deep learning technology, especially the convolutional neural network (CNN) has been widely used in many fields such as Computer Vision (CV), Natural Language Processing (NLP) and Automatic Speech Recognition (ASR). CNNs have become the state-of-the-art technique in many vision tasks, such as image classification, object detection, etc. But the deep CNNs may make part of the kernels too thin by using parameterized convolution kernel to extract features. Therefore, this paper proposes a method to optimize CNNs by calculating the similarity coefficient between the feature maps. Experimental results showed that this method improved the training speed and the detecting speed with the accuracy been ensured.