Journal Home Online First Current Issue Archive For Authors Journal Information 中文版

Strategic Study of CAE >> 2005, Volume 7, Issue 10

Algorithm Design for Improving Feature Extraction Efficiency Based on KPCA

Department of Computer Science & Technology, Nanjing University of Science & Technology, Nanjing 210094, China

Funding project:国家自然科学基金资助项目(60072034) Received: 2004-09-21 Revised: 2004-11-10 Available online: 2005-10-20

Next Previous

Abstract

KPCA (kernel PCA) is derived from PCA. It can extract nonlinear feature components of samples. However, feature extraction for one sample requires that kernel functions between training samples and the sample be calculated in advance. So, the size of training sample set affects the efficiency of feature extraction. It is supposed that in feature space the eigenvectors may be linearly expressed by a part of training samples, called nodes. According to the supposition, an improved KPCA (IKPCA) algorithm is developed. IKPCA extracts feature components of one sample efficiently, only based on kernel functions between nodes and the sample. Experimental results show that IKPCA is very close to KPCA in performance, while with higher efficiency.

References

[ 1 ] 边肇祺, 张学工.模式识别[M].北京:清华大学出版社, 2000 link1

[ 2 ] DudaRO, HartPE, StorkDG.模式分类[M].北京:机械工业出版社, 中信出版社, 2003 link1

[ 3 ] ScholkopfB, SmolaA, M櫣llerKR.Nonlinear componentanalysisasakerneleigenvalueproblem[J].NeuralComputation, 1998, 10 (5) :1299~1319

[ 4 ] MikaS, R tschG, WestonJ, Sch lkopfB, M櫣llerK R.Fisherdiscriminateanalysiswithkernels[A].In:HuYH, LarsenJ, WilsonE, DouglasS, eds.Neural NetworksforSignalProcessingIX, IEEE[C].1999.41~48

[ 5 ] MikaS, SmolaAJ, Sch lkopfB.Animprovedtrainingalgorithmforkernelfisherdiscriminants[A].In:JaakkolaT, RichardsonT, eds.Proceedings AISTATS[C].MorganKaufmann, 2001.98~104

[ 6 ] 徐 勇, 杨静宇, 金 忠, 娄 震.一种基于核的快速非线性鉴别分析方法[J].计算机研究与发展, 2004, (1) link1

[ 7 ] BillingsSA, LeeKL.Nonlinearfisherdiscriminant analysisusingaminimumsquarederrorcostfunction andtheorthogonalleastsquaresalgorithm[J].Neural Networks, 2002, 15 (2) :263~270

[ 8 ] XuJ, ZhangX, LiY.KernelMSEalgorithm:a unifiedframeworkforKFD, LS SVMandKRR[A].In:ProceedingsoftheInternationalJointConference onNeuralNetworks (IJCNN2001) [C].Washington, DC, 2001.1486~1491

[ 9 ] 福永圭之介.统计图形识别导论[M].陶笃纯译.北京:科学出版社, 1978 link1

[10] 金 忠.人脸图像特征抽取与维数研究[D].南京:南京理工大学, 1999

[11] XuYong, YangJingyu, YangJian, Areformative kernelFisherdiscriminantanalysis[J].Pattern Recognition, 2004, 37:1299~1302

[12] XuYong, YangJingyu, LuJianfeng, YuDongjun, AnefficientrenovationonkernelFisherdiscriminant analysisandfacerecognitionexperiments[J].Pattern Recognition, 2004, 37:2091~2094

Related Research