Browse > Article
http://dx.doi.org/10.5391/IJFIS.2003.3.1.100

One-Class Support Vector Learning and Linear Matrix Inequalities  

Park, Jooyoung (Dept. of Control ＆ Instrumentation Engineering, Korea University)
Kim, Jinsung (Dept. of Electrical Engineering, Korea University)
Lee, Hansung (Dept. of Computer and Information Science, Korea University)
Park, Daihee (Dept. of Computer and Information Science, Korea University)
Publication Information
International Journal of Fuzzy Logic and Intelligent Systems / v.3, no.1, 2003 , pp. 100-104 More about this Journal
Abstract
The SVDD(support vector data description) is one of the most well-known one-class support vector learning methods, in which one tries the strategy of utilizing balls defined on the kernel feature space in order to distinguish a set of normal data from all other possible abnormal objects. The major concern of this paper is to consider the problem of modifying the SVDD into the direction of utilizing ellipsoids instead of balls in order to enable better classification performance. After a brief review about the original SVDD method, this paper establishes a new method utilizing ellipsoids in feature space, and presents a solution in the form of SDP(semi-definite programming) which is an optimization problem based on linear matrix inequalities.
Keywords
One-class classification; Support vector learning; Ellipsoid; Linear matrix inequality; Semi-definite programming;
Citations & Related Records
연도 인용수 순위
  • Reference
1 D. Tax and R. Duin, 'Support Vector DomainDescription,' Pattern Recognition Letters, vol. 20, pp.1191-1199, 1999   DOI   ScienceOn
2 C. Campbell and K. P. Bennett, 'A linear programmingapproach to novelty detection,' Advances of NIPS 2000,pp. 395-401, 2000
3 P. Gahinet, A. Nemirovski, A. J. Laub and M. Chilali, LMI control toolbox, Math Works Inc., Natick, MA,1995
4 G. Ratch, S. Mika, B. Sch$\ddot{o}$lkopf, and K.-R. M$\ddot{u}$ller,'Constructing boosting algorithms from SVMs: Anapplication to one-class classification,' IEEETransactions on Pattern AnaIysis and MachineIntelligence, vol. 24, pp. 1-15, 2002   DOI   ScienceOn
5 B. Sch$\ddot{o}$lkopf and A. J. Smola, Learning with kernels,MIT Press, 2002
6 S. Boyd, L. ElGhaoui, E. Feron and V. Balakrishnan, Linear matrix inequalities in systems and control theory, SIAM Studies in Apptied Mathematics, Vot. 15, SIAM,Philadelphia, 1994
7 D. Tax and P. Juszczak, 'Kemel whitening forone-class classification,' Pattern Recognition with Support Vector Machines, pp. 40-52, 2002
8 N. Cristianinian J. Shawe-Taylor, An introduction tosupport vector machines and other kernel-based learningmethods, Cambridge University Press, 2000
9 K. Tsuda, 'Support vector classifiers with asymmetrickernel functions,' Proceedings of ESANN, PP. 183-188,1999
10 D. Tax, One-class classification, PhD Thesis, DelftUniversity of Technology, 2001
11 B. Scholkopf, J. C. Platt, and A. J. Smola, Kernetmethod for percentite feature extraction, Technical Report MSR-TR-2000-22, Microso A Research, WA,2000
12 B. Sch$\ddot{o}$lkopf, J. C. Platt, J. Shawe-TayIor, and A. J.Smola, and R. C. Williamson, 'Estimating the supportof a high-dimensional distribution,' Neural Computation,vol.13, pp. 1443-1471, 2001   DOI   ScienceOn
13 C. BishoP, 'NoveIty detection and neural networksvalidation, IEE Proceedings on Vision, Image, and Signal Processing, Special Issue on Apptications of Neural Networks, vol. 141, pp. 217-222, 1994