论文信息 - Fast training Support Vector Machines using parallel sequential minimal optimization

Fast training Support Vector Machines using parallel sequential minimal optimization

One of the key factors that limit support vector machines (SVMs) application in large sample problems is that the large-scale quadratic programming (QP) that arises from SVMs training cannot be easily solved via standard QP technique. The sequential minimal optimization (SMO) is current one of the major methods for solving SVMs. This method, to a certain extent, can decrease the degree of difficulty of a QP problem through decomposition strategies, however, the high training price for saving memory space must be endured. In this paper, an algorithm in the light of the idea of parallel computing based on Symmetric multiprocessor (SMP) machine is improved. The new technique has great advantage in terms of speediness when applied to problems with large training sets and high dimensional spaces without reducing generalization performance of SVMs. .

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

[2] Hans Ulrich Simon,et al. A General Convergence Theorem for the Decomposition Method , 2004, COLT.

[3] Chih-Jen Lin,et al. Asymptotic convergence of an SMO algorithm without any assumptions , 2002, IEEE Trans. Neural Networks.

[4] S. Sathiya Keerthi,et al. Improvements to Platt's SMO Algorithm for SVM Classifier Design , 2001, Neural Computation.

[6] Chih-Jen Lin,et al. A Study on SMO-Type Decomposition Methods for Support Vector Machines , 2006, IEEE Transactions on Neural Networks.

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

[8] Chih-Jen Lin,et al. On the convergence of the decomposition method for support vector machines , 2001, IEEE Trans. Neural Networks.

[9] Chih-Jen Lin,et al. A formal analysis of stopping criteria of decomposition methods for support vector machines , 2002, IEEE Trans. Neural Networks.