[1]余正涛,邹俊杰,赵兴,等.基于主动学习的最小二乘支持向量机稀疏化[J].南京理工大学学报(自然科学版),2012,36(01):12-17.
 YU Zheng-tao,ZOU Jun-jie,ZHAO Xing,et al.Sparseness of Least Squares Support Vector Machines Based on Active Learning[J].Journal of Nanjing University of Science and Technology,2012,36(01):12-17.
点击复制

基于主动学习的最小二乘支持向量机稀疏化
分享到:

《南京理工大学学报》(自然科学版)[ISSN:1005-9830/CN:32-1397/N]

卷:
36卷
期数:
2012年01期
页码:
12-17
栏目:
出版日期:
2012-02-29

文章信息/Info

Title:
Sparseness of Least Squares Support Vector Machines Based on Active Learning
作者:
余正涛; 邹俊杰; 赵兴; 苏磊; 毛存礼;
昆明理工大学信息工程与自动化学院; 昆明理工大学智能信息处理重点实验室;
Author(s):
YU Zheng-tao12ZOU Jun-jie12ZHAO Xing12SU Lei12MAO Cun-li12
1.School of Information Engineering and Automation;2.Key Laboratory of Intelligent Information Processing, Kunming University of Science and Technology,Kunming 650051,China
关键词:
最小二乘支持向量机 稀疏化 主动学习 分类
Keywords:
least squares support vector machines sparseness active learning classification
分类号:
TP181
摘要:
针对最小二乘支持向量机(LSSVM)稀疏化问题,提出一种基于主动学习的LSSVM数据稀疏化学习算法。首先基于核聚类的方法选取初始样本,并利用LSSVM构建一个最小分类器,然后计算样本在分类器作用下的分布,选择最接近分类面的样本进行标记,最后将该标记样本加入训练集建立新的分类器,重复上述过程直到模型精度满足要求,以此建立部分样本的LSSVM稀疏化模型。利用加利福尼亚大学欧文分校(UCI)提供的6种数据集进行实验,结果表明,提出的方法使LSSVM的稀疏性提高了46%以上,减少了标注样本带来的成本。
Abstract:
To solve the sparseness problem of least squares support vector machine(LSSVM)learning process,this paper proposes a learning algorithm of LSSVM data sparseness based on active learning.This algorithm first selects initial samples based on a kernel clustering method and constructs a minimum classification using LSSVM,calculates the sample distribution under the action of the classifier,and labels the samples closest to hyper planes.These labeled samples are finally added into the training sets to train a new classifier,and the processes are repeated until the model accuracy meets requirements.The LSSVM sparse model of some samples are established.Experiments on the University of California Irvine(UCI)data sets show that the proposed algorithm can increase the sparseness of LSSVM by more 46 percent and reduce the cost of labeling samples

参考文献/References:

[1] Dong Jiaxiong,Krzyzak A,Suen C Y. Fast SVM training algorithm with decomposition on very large data sets[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2005, 27( 4) : 603-618.
[2] Joachims T. Making large scale SVM learning practical [M]. Boston: MIT Press, 1999: 169-184.
[3] Ji Aibing,Pang Jiahong,Li Shuhuan, Sun Jianpin. Support vector machine for classification based on fuzzy training data[A]. International Conference on Machine Learning and Cybernetics[C]. Dalian,China: IEEE Press,2006: 1609-1614.
[4] 周晓剑,马义中,朱嘉钢. SMO 算法的简化及其在非正定核条件下的应用[J]. 计算机研究与发展. 2010, 47( 11) : 1962-1969.
[5] Catanzaro B C,Sundaram N,Keutzer K. Fast support vector machine training and classification on graphics processors[A]. International Conference on Machine Learning[C]. Helsinki,Finland: ACM Press,2008: 104-111.
[6] 张玉珍,何新,王建宇,等. 一种基于SVM 的高效球门检测方法[J]. 南京理工大学学报,2010,34( 1) : 13-18.
[7] Suykens J A K,Vandewalle J. Least squares support vector machine classifiers [J]. Neural Processing Letters, 1998,9 ( 3) : 293-300.
[8] Suykens J A K,De Brabanter J,Lukas L, et al. Weighted least squares support vector machines: robustness and sparse approximation[J]. Neuro Computing, 2002, 48( 1) : 85-105.
[9] Freund Y,Seung H,Shamir E, et al. Selective sampling using the query by committee algorithm[J]. Machine Learning, 1997, 28( 2) : 133-168. .[10] Kenji FukumiZu. Statistical active learning in multilayer perceptrons[J]. IEEE Transactions on Neural Networks, 2000, 11( 1) : 17-26.
[11] LindenBaum M,Markovitch S,RusaKov D. Selective sampling for nearest neighbor classifiers[J]. Machine Learning, 2004, 54( 2) : 125-152.
[12] Simon H A,Lea G. Problem solving and rule education: A unified view knowledge and organization[J]. Erbuam, 1974, 15( 2) : 63-73.
[13] Li M,Sethi I K. Confidence-based active learning[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2006, 28( 8) : 1251-1261.
[14] Hoegaerts L, Suykens J A K,Vandewallej, et al. A comparison of pruning algorithms for sparse least squares support vector machines[A]. Proceedings of the 11th International Conference on Neural Information Processing[C]. Calcutta, India: IOA Press, 2004, 3316: 1247-1253.
[15] 陶少辉,陈德钊,胡望明. 基于CCA 对LSSVM 分类器的稀疏化[J]. 浙江大学学报( 工学版) ,2007, 41( 7) : 1093-1096, 1118.
[16] Tong S,Chang E. Support vector machine active learning for image retrieval[A]. Proceedings of the 9th ACM International Conference on Multimedia [C]. Ottawa, Canada: ACM Press, 2001: 107-118.
[17] Michael I M,Graham E P,Daniel P E. Support vector machine active learning for music retrieval[J]. Multimedia Systems, 2006, 12( 1) : 3-13.
[18] Tong S,Koller D. Support vector machine active learning with applications to text classification[J]. Journal of Machine Learning Research, 2002. 2( 3) : 45-66.
[19] University of California Irvine. http: / /archive. ics. uci. edu /ml /index. html[EB/OL]. 2011-05-01.
[20] University of Waikato. http: / /www. cs. waikato. ac. nz / ml /weka[EB/OL]. 2011-06-30.
[21] Kris D B. http: / /www. esat. kuleuven. be/sista/lssvmlab [EB/OL]. 2011-05-15.

备注/Memo

备注/Memo:
国家自然科学基金(60863011,61175028);云南省自然科学基金重点项目(2008CC023);云南省中青年学术和技术带头人后备人才项目(2007PY01-11)
更新日期/Last Update: 2012-10-12