一种基于QR分解的增量式核判别分析法

计算机科学 ›› 2014, Vol. 41 ›› Issue (4): 297-301.

一种基于QR分解的增量式核判别分析法

王万良,陈宇,邱虹,郑建炜

浙江工业大学计算机学院杭州310023;浙江工业大学计算机学院杭州310023;浙江工业大学计算机学院杭州310023;浙江工业大学计算机学院杭州310023

出版日期:2018-11-14 发布日期:2018-11-14
基金资助:
本文受国家科技支撑计划课题(2012BAD10B01),浙江省自然科学基金(LQ12F03011)资助

Incremental Kernel Discriminant Analysis Method via QR Decomposition

WANG Wan-liang,CHEN Yu,QIU Hong and ZHENG Jian-wei

Online:2018-11-14 Published:2018-11-14

摘要/Abstract

摘要： 针对非线性系统在线学习的效率问题,提出了一种基于QR分解的增量式核判别分析法。该算法充分利用基于QR分解的核判别分析法的先降维后提取特征的思想,将核空间映射到低维空间进行计算,减少了构造核矩阵的计算量,降低了核矩阵的存储空间。同时引入增量计算的思想,有效地解决了在线学习中冗余计算的问题。在TE过程数据和ORL人脸库上的仿真实验证明了该算法在特征提取上的有效性,其相比批量式算法有更高的效率优势。

关键词: 非线性系统,在线学习,QR分解,增量计算,特征降维

Abstract: To improve the online learning efficiency of nonlinear system,an incremental kernel decomposition analysis method via QR decomposition was developed．This algorithm maps the kernel space to a lower dimension space before performing feature decomposition to reduce the amount of calculation and the storage space of the kernel matrix．Then this algorithm solves the problem of redundant computation by combining the idea of incremental computation．The experiments on TE process and ORL data set show this algorithm is effective on feature decomposition as well as more efficient than the batch one.

Key words: Nonlinear system,Online learning,QR decomposition,Incremental computation,Dimension reduction

王万良,陈宇,邱虹,郑建炜. 一种基于QR分解的增量式核判别分析法[J]. 计算机科学, 2014, 41(4): 297-301. https://doi.org/

WANG Wan-liang,CHEN Yu,QIU Hong and ZHENG Jian-wei. Incremental Kernel Discriminant Analysis Method via QR Decomposition[J]. Computer Science, 2014, 41(4): 297-301. https://doi.org/

参考文献

[1] Peng H,Li B,Luo X,et al．A learning-based audio watermarking scheme using kernel Fisher discriminant analysis [J]．Digital Signal Processing,2013,23(1):382-389
[2] Meng Y,Zou J,Gan X,et al．Adaptive WNN aerodynamic mo-deling based on subset KPCA feature extraction [J]．Journal of Central South University,2013,20:931-941
[3] 胡益,王丽,马贺贺,等．基于核 PLS 方法的非线性过程在线监控[J]．化工学报,2011,62(9):2555-2561
[4] Gan J Y,Liu J F．Fusion and recognition of face and iris feature based on wavelet feature and KFDA [C]∥ International Confe-rence on Wavelet Analysis and Pattern Recognition,2009．ICWAPR 2009．IEEE,2009:47-50
[5] Zhang X,Yan W,Zhao X,et al．Nonlinear biological batch process monitoring and fault identification based on kernel fisher discriminant analysis [J]．Process biochemistry,2007,42(8):1200-1210
[6] 郑建炜,王万良,姚晓敏,等．张量局部Fisher判别分析的人脸识别[J]．自动化学报,2012,38(9):1485-1495
[7] Jin L W,Ding K,Huang Z B．Incremental learning of FDA mo-del for Chinese writer adaptation [J]．Neurocomputing,2010,73(10):1614-1623
[8] Xiong T,Ye J,Li Q,et al．Efficient kernel discriminant analysis via QR decomposition [J]．Advances in neural information processing systems,2005,17:1529-1536
[9] Golub G H,Van Loan C F．Matrix computations [M]．JohnsHopkins University Press,1996
[10] Zhang X,Yan W W,Zhao X,et al．Performance monitoring and fault identification based on DKPCA and PSVM [J]．Journal of Shanghai Jiaotong University,2008,42(2):181-185
[11] Chiang L H,Russell E,Braatz R D．Fault detection and diagnosis in industrial systems [M]．Springer Verlag,2001
[12] Bo C M,Zhang S,Zhang G M,et al．Fault identification of Tennessee Eastman process based on FS-KPCA [J]．Journal of Chemical Industry and Engineering,2008,59(7):1783-1789
[13] Ricker N L,Lee J H．Nonlinear model predictive control of the Tennessee Eastman challenge process [J]．Computers & chemical engineering,1995,19(9):961-981
[14] 张熠卓,徐光华,梁霖,等．利用增量式非线性流形学习的状态监测方法[J]．西安交通大学学报,2011,45(001):64-68
[15] 王迎,王新明,赵小强．基于小波去噪与 KPCA 的 TE 过程故障检测研究[J]．化工机械,2011,38(1):49-53
[16] http://blog.sina.com.cn/s/blog_02b6f23d0100eq88.html

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed