摘要
本文首先介绍卷积神经网络深度模型的发展历程,其次对其的内部组成结构所对应的各项操作(卷积、池化)进行了详细的阐述分析,最后较全面地分析卷积神经网络在语音识别、表情检测、遥感图像分类及其他领域的具体应用。通过总结研究卷积神经网络技术热点问题,研究人员能够从中提出有用的意见同时发掘出更好的改进方法。
This article first introduces the development of convolution neural network depth model.Secondly,the various operations(convolution and pooling)corresponding to its internal composition are elaborated and analyzed in detail.Finally,a more comprehensive analysis of convolution neural network in speech recognition,facial expression detection,remote sensing image classification and other specific applications.By summarizing the hot issues in convolution neural network technology,researchers can come up with useful ideas and discover better methods to improve it.
作者
刘方园
王水花
张煜东
LIU Fang-yuan;WANG Shui-hua;ZHANG Yu-dong(School of Computer Science and Technology,Nanjing Normal University,Nanjing 210023,China)
出处
《新型工业化》
2017年第11期40-51,共12页
The Journal of New Industrialization
基金
国家重点研发计划(2017YFB1103200/02)
国家自然科学基金(61602250
61503188)
江苏省自然科学基金(BK20150983
BK20150982)
江苏省高校自然科学研究面上项目(16KJB520025
15KJB470010)
关键词
卷积神经网络
语音识别
人脸表情
遥感图像
Convolution Neural Network
Speech recognition
Facial expression
Remote sensing image