[1]陈浩翔,蔡建明,刘铿然,等. 手写数字深度特征学习与识别[J].计算机技术与发展,2016,26(07):19-2.
　CHEN Hao-xiang,CAI Jian-ming,LIU Keng-ran,et al. Deep Learning and Recognition of Handwritten Numeral Features[J].,2016,26(07):19-2.
点击复制

 手写数字深度特征学习与识别()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
26
期数:
2016年07期
页码:
19-2
栏目:
智能、算法、系统工程
出版日期:
2016-07-10

文章信息/Info

Title:
 Deep Learning and Recognition of Handwritten Numeral Features
文章编号:
1673-629X（2016)07-0019-05
作者:
 陈浩翔; 蔡建明; 刘铿然; 林秋爽; 张文玲; 周涛
 华南师范大学 数学科学学院
Author(s):
 CHEN Hao-xiang; CAI Jian-ming; LIU Keng-ran; LIN Qiu-shuang; ZHANG Wen-ling; ZHOU Tao
关键词:
 ; 深度学习; 特征融合; 特征提取; 手写数字识别; 主成分分析; 梯度方向直方图
Keywords:
 ; deep learning; feature fusion; feature extraction; handwritten numeral recognition; principal component analysis ; histogram of oriented gradient
分类号:
O235
文献标志码:
A
摘要:
 深度学习中的网络结构设计、特征提取与融合是数据挖掘和模式识别理论和行业应用中的关键问题。文中以相关领域中的典型应用问题手写数字识别和权威数据库MNIST为实验平台(包含七万个手写数字图像),探索了深度学习网络结构的设计和特征融合问题,保证研究结果的实用性、代表性和可参考性。所给方案的步骤是：首先,设计非监督深度学习网络,进行非监督高层语义特征学习,提取深度特征( DF),探索特征的高层认知特点；其次,对手写数字数据库进行非监督多特征提取,包括HOG(梯度方向直方图)特征、PCA(主成分分析)特征、LDA(判别分析)特征、像素分布特征、穿越次数特征和投影特征,构建手写数字典型特征库( Library of Typical Features,LTF)；最后,构建深度有监督学习网络,有监督地融合深度特征DF和典型特征库。实验结果表明,相比于文献中的典型特征,该方案能够将手写数字识别的错误率有效降低50%。
Abstract:
 Network structure design,feature extraction and fusion in deep learning are key problems in data mining and pattern recognition theory and industry application. The design of deep learning network’s structure and the problem of feature fusion is explored,taking handwritten numeral recognition and authoritative database MNIST,with 70 thousands of handwritten image,as the experiment platform, which guarantees the practicability,representation and reference of the research results. The solution step has been given. Firstly,the unsu-pervised deep learning network is designed,learning unsupervised high-level semantic features,extraction of depth features,and explora-tion of higher cognitive characteristics of features. Secondly,unsupervised features of handwritten database are extracted,including HOG, PCA,LDA and so on,construction of LTF. Finally,deep supervised learning network is built,fusion of deep features and the library of typical features with supervision. The result shows that this scheme can lower error rate of handwritten recognition by 50%,compared with the typical features of the present.

相似文献/References:

[1]张志宏,吴庆波,邵立松,等.基于飞腾平台TOE协议栈的设计与实现[J].计算机技术与发展,2014,24(07):1.
　ZHANG Zhi-hong,WU Qing-bo,SHAO Li-song,et al. Design and Implementation of TCP/IP Offload Engine Protocol Stack Based on FT Platform[J].,2014,24(07):1.
[2]梁文快,李毅. 改进的基因表达算法对航班优化排序问题研究[J].计算机技术与发展,2014,24(07):5.
　LIANG Wen-kuai,LI Yi. Research on Optimization of Flight Scheduling Problem Based on Improved Gene Expression Algorithm[J].,2014,24(07):5.
[3]黄静,王枫,谢志新,等. EAST文档管理系统的设计与实现[J].计算机技术与发展,2014,24(07):13.
　HUANG Jing,WANG Feng,XIE Zhi-xin,et al. Design and Implementation of EAST Document Management System[J].,2014,24(07):13.
[4]侯善江[],张代远[][][]. 基于样条权函数神经网络P2P流量识别方法[J].计算机技术与发展,2014,24(07):21.
　HOU Shan-jiang[],ZHANG Dai-yuan[][][]. P2P Traffic Identification Based on Spline Weight Function Neural Network[J].,2014,24(07):21.
[5]李璨,耿国华,李康,等. 一种基于三维模型的文物碎片线图生成方法[J].计算机技术与发展,2014,24(07):25.
　LI Can,GENG Guo-hua,LI Kang,et al. A Method of Obtaining Cultural Debris’ s Line Chart Based on Three-dimensional Model[J].,2014,24(07):25.
[6]翁鹤,皮德常. 混沌RBF神经网络异常检测算法[J].计算机技术与发展,2014,24(07):29.
　WENG He,PI De-chang. Chaotic RBF Neural Network Anomaly Detection Algorithm[J].,2014,24(07):29.
[7]刘茜[],荆晓远[],李文倩[],等. 基于流形学习的正交稀疏保留投影[J].计算机技术与发展,2014,24(07):34.
　LIU Qian[],JING Xiao-yuan[,LI Wen-qian[],et al. Orthogonal Sparsity Preserving Projections Based on Manifold Learning[J].,2014,24(07):34.
[8]尚福华,李想,巩淼. 基于模糊框架-产生式知识表示及推理研究[J].计算机技术与发展,2014,24(07):38.
　SHANG Fu-hua,LI Xiang,GONG Miao. Research on Knowledge Representation and Inference Based on Fuzzy Framework-production[J].,2014,24(07):38.
[9]叶偲,李良福,肖樟树. 一种去除运动目标重影的图像镶嵌方法研究[J].计算机技术与发展,2014,24(07):43.
　YE Si,LI Liang-fu,XIAO Zhang-shu. Research of an Image Mosaic Method for Removing Ghost of Moving Targets[J].,2014,24(07):43.
[10]余松平[][],蔡志平[],吴建进[],等. GSM-R信令监测选择录音系统设计与实现[J].计算机技术与发展,2014,24(07):47.
　YU Song-ping[][],CAI Zhi-ping[] WU Jian-jin[],GU Feng-zhi[]. Design and Implementation of an Optional Voice Recording System Based on GSM-R Signaling Monitoring[J].,2014,24(07):47.
[11]贺飞翔,赵启军. 基于深度学习的头部姿态估计[J].计算机技术与发展,2016,26(11):1.
　HE Fei-xiang,ZHAO Qi-jun. Head Pose Estimation Based on Deep Learning[J].,2016,26(07):1.
[12]曾志平[] [],萧海东[],张新鹏[]. 基于DBN的金融时序数据建模与决策[J].计算机技术与发展,2017,27(04):1.
　ZENG Zhi-ping[] [],XIAO Hai-dong[],ZHANG Xin-peng[]. Modeling and Decision-making of Financial Time Series Data with DBN[J].,2017,27(07):1.
[13]李宏林. 分析式纹理合成技术及其在深度学习的应用[J].计算机技术与发展,2017,27(11):7.
　LI Hong-lin. Analyzed Texture-synthesis Techniques and Their Applications in Deep Learning[J].,2017,27(07):7.

更新日期/Last Update: 2016-09-28