[1]潘志松,燕继坤,杨绪兵,等.单实例分类算法研究[J].南京理工大学学报(自然科学版),2009,(04):444-449.
 PAN Zhi-song,YAN Ji-kun,YANG Xu-bing,et al.Classification Algorithm Based on Single Sample[J].Journal of Nanjing University of Science and Technology,2009,(04):444-449.
点击复制

单实例分类算法研究
分享到:

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

卷:
期数:
2009年04期
页码:
444-449
栏目:
出版日期:
2009-08-30

文章信息/Info

Title:
Classification Algorithm Based on Single Sample
作者:
潘志松 1 燕继坤 2 杨绪兵 3 缪志敏 1 陈 斌 3
1.解放军理工大学指挥自动化学院, 江苏南京210007; 2.西南电子研究所,四川成都610041; 3.南京航空航天大学计算机科学与技术学院,江苏南京210016
Author(s):
PAN Zhi-song1YAN Ji-kun2YANG Xu-bing3MIAO Zhi-min1CHEN Bin3
1.Institute of Command Automation,PLA University of Science and Technology,Nanjing 210007,China;2.The West-South Electronics Institute,Chengdu 610041,China;3.Department of ComputerScience and Engineering,Nanjing University of Aeronautics & Astronautics,Na
关键词:
单实例 核方法 分类 支持向量
Keywords:
single samples kernel means classification support vectors
分类号:
TP18
摘要:
针对不平衡分类问题的极端情况,即用于训练的样本极少甚至只有一个实例,该文提出了一种单实例分类算法,这种方法使用球面作为分类面,在目标类的单实例在球内和反类尽量位于球面外的约束条件下,最大化该分类球面的半径,该方法能够有效地处理线性可分的数据分布。当输入样本分布结构呈高度非线性时,该算法通过核映射将低维输入空间中的非线性可分问题变换为高维特征空间中可能的线性可分问题,并以内积形式刻画,最终在特征空间上通过核技巧获得原问题的解决。通过对标准数据集和实际数据集的实验,验证了单实例分类算法在处理数据不平衡问题上的有效性。
Abstract:
In order to solve the extreme situation that only a few target examples or only one can be used in training the classification,a single sample classification algorithm is presented here.Spherical surfaces are applied as classified hypersphere,and the largest radius can be obtained enclosing the single sample under the restriction that all outliers are outside the hypersphere.It fails when the distribution of input patterns is complex.The classifier applies kernel means,performing a nonlinear data transformation into some high dimensional feature space,increases the probability of the linear separability of the patterns within the feature space and therefore solves the original classification problem.The paper verifies that the algorithm can effectively deal with the unbalanced data classification on various synthetic and UCI datasets.

参考文献/References:

[ 1] Mitchell T. Machine learning[M]. NewYork, USA: McGraw-HillCompanies, 1997.
[ 2] Moya M, KochM, Hostetler L. One-class classifer networks for target recognitionapplications[A]. Pro-ceedings ofWorldCongress onNeuralNetworks[ C]. Oregen, Portland: InternationalNeuralNetworkSocie-ty, 1993. 797-801.
[ 3] TaxD. One-classclassification-concept-learning inthe absence of counter-examples[ D]. Delft, Holland: DelftUniversity ofTechnology, 2001.
[ 4] Tax D, DuinR. Support vector domain description [ J]. Pattern Recognition Letters, 1999, 20( 11-13): 1191- 1199.
[ 5] VapnikVN. The nature of statistical learning theory [M]. Berlin, Germany: Springer-Verlag, 1999.
[ 6] CristianiniN, Taylor J. An introductionto SVMs and other kerne-lbased learning methods[M]. London, UK: CambridgeUniv Press, 2000.
[ 7] Sch’ olkopf B, WilliansonR, SmolaA, et a.l Support vectormethod for novelty detection[ J]. Advances in Neural Information Processing Systems, 1999, 12: 582-588.
[ 8] Juszczak P. Learning to recognize) Study on one-class classifcation and active learning [ D]. Delft, Holland: DelftUniversity ofTechnology, 2006.
[ 9] ScholkopfB, BurgesC, VapnikV. Extracting support data for a given task[A]. First International Confer-ence on Knowledge Discovery& Data Mining[ C]. Menlo Park, CA: AAAI Press, 1995. 252-257.
[ 10] BlakeC, MerzC. UCI repository ofmachine learning databases[ EB/OL]. http: //www. ics. uc.i edu/m-l earn/MLRepository. htm,l 1998.

备注/Memo

备注/Memo:
基金项目: 国家自然科学基金( 60603029)
作者简介: 潘志松( 1973- ), 男,博士, 副教授,主要研究方向: 模式识别,网络安全, E-mail:Hotpzs@hotmail.com。
更新日期/Last Update: 2012-11-19