Incremental training of support vector machines using hyperspheres

In the conventional incremental training of support vector machines, candidates for support vectors tend to be deleted if the separating hyperplane rotates as the training data are added. To solve this problem, in this paper, we propose an incremental training method using one-class support vector machines. First, we generate a hypersphere for each class. Then, we keep data that exist near the boundary of the hypersphere as candidates for support vectors and delete others. By computer simulations for two-class and multiclass benchmark data sets, we show that we can delete data considerably without deteriorating the generalization ability.

[1]  阿部 重夫.  Pattern classification : neuro-fuzzy methods and their comparison , 2001 .

[2]  Gert Cauwenberghs,et al.  Incremental and Decremental Support Vector Machine Learning , 2000, NIPS.

[3]  Thomas G. Dietterich,et al.  Solving Multiclass Learning Problems via Error-Correcting Output Codes , 1994, J. Artif. Intell. Res..

[4]  James M. Keller,et al.  Will the real iris data please stand up? , 1999, IEEE Trans. Fuzzy Syst..

[5]  Shigeo Abe,et al.  Input Layer Optimization of Neural Networks by Sensitivity Analysis and its Application to Recognition of Numerals , 1991 .

[6]  C. A. Murthy,et al.  Data condensation in large databases by incremental learning with support vector machines , 2000, Proceedings 15th International Conference on Pattern Recognition. ICPR-2000.

[7]  Yoshua Bengio,et al.  Gradient-based learning applied to document recognition , 1998, Proc. IEEE.

[8]  Dimitrios Gunopulos,et al.  Incremental support vector machine construction , 2001, Proceedings 2001 IEEE International Conference on Data Mining.

[9]  Liva Ralaivola,et al.  Incremental Support Vector Machine Learning: A Local Approach , 2001, ICANN.

[10]  Noboru Murata,et al.  Optimisation on support vector machines , 2000, Proceedings of the IEEE-INNS-ENNS International Joint Conference on Neural Networks. IJCNN 2000. Neural Computing: New Challenges and Perspectives for the New Millennium.

[11]  R. Fisher.  THE USE OF MULTIPLE MEASUREMENTS IN TAXONOMIC PROBLEMS , 1936 .

[13]  Vladimir N. Vapnik,et al.  The Nature of Statistical Learning Theory , 2000, Statistics for Engineering and Information Science.

[14]  Vladimir Vapnik,et al.  Statistical learning theory , 1998 .

[15]  João Pedro Pedroso,et al.  Optimization on Support Vector Machines. , 2000 .

[16]  Sholom M. Weiss,et al.  An Empirical Comparison of Pattern Recognition, Neural Nets, and Machine Learning Classification Methods , 1989, IJCAI.

[17]  Rong Xiao,et al.  An Approach to Incremental SVM Learning Algorithm , 2000, 2008 ISECS International Colloquium on Computing, Communication, Control, and Management.

[18]  Marimuthu Palaniswami,et al.  Incremental training of support vector machines , 2005, IEEE Transactions on Neural Networks.

[19]  Robert P. W. Duin,et al.  Outliers and data descriptions , 2001 .

[20]  Ulrich H.-G. Kreßel,et al.  Pairwise classification and support vector machines , 1999 .