A new nu-support vector machine for training sets with duplicate samples

Analyzed theoretically, /spl nu/-SVM was found to be over-dependent on each training sample, even if the samples have same value. This dependence would result in more time for training, more support vectors and more decision time. In order to overcome this problem, we propose a new /spl nu/-SVM. This new /spl nu/-SVM multiplies each slack variable in the objective function by a weight factor, and automatically computes each weight factor by the number of corresponding samples with same value before training. Theoretical analysis and the results of experiments show that the new /spl nu/-SVM has the same classification precision rate as the standard /spl nu/-SVM and the new /spl nu/-SVM is faster than the /spl nu/-SVM in training and decision if the training sets have same value samples.

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

[2]  John C. Platt,et al.  Fast training of support vector machines using sequential minimal optimization, advances in kernel methods , 1999 .

[3]  Corinna Cortes,et al.  Support-Vector Networks , 1995, Machine Learning.

[4]  Cheng-Chew Lim,et al.  Dual /spl nu/-support vector machine with error rate and training size biasing , 2001, 2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.01CH37221).

[5]  Bernhard Schölkopf,et al.  New Support Vector Algorithms , 2000, Neural Computation.

[6]  H. P. Huang,et al.  Fuzzy Support Vector Machines for Pattern Recognition and Data Mining , 2002 .

[7]  R. Fletcher.  Practical Methods of Optimization , 1988 .

[8]  Alexander J. Smola,et al.  Learning with kernels , 1998 .