近年来,深度学习作为机器学习的新兴研究领域越来越受到人们的关注,通过深度学习构建的深度网络在无监督特征提取方面表现出优异性能. 卷积神经网络作为一个相对成功的深度学习模型,逐渐成为模式识别领域的研究热点. 本文对卷积神经网络及其近年来在模式识别领域取得的新进展进行综述. 首先介绍深度学习与卷积神经网络之间的关系以及卷积神经网络的基本原理;其次对卷积神经网络的各种改进算法进行了总结,对卷积神经网络在模式识别领域的新应用进行了概述;最后阐述了目前在卷积神经网络学习理论中亟需解决的主要问题.%Recently,deep learning has attracted more and more attention as a new research area in machine learning.The deep net-work model constructed from deep learning shows the excellent performance in unsupervised feature extraction.Convolutional neural network ( CNN) is a relative successful deep learning model and it has gradually become the focus of current research. A general progress overview for CNN and its new progress in pattern recognition are gave.The basic situation of deep learning and CNN are in-troduced at first,including the fundamental principles of CNN,the relationship about deep learning and CNN.Then,the improved al-gorithms about CNN and its new applications in pattern recognition are summarized.Finally,the issues about CNN need to solve in the future are discussed.
展开▼