Title :
Global convergence analysis of decomposition methods for support vector machines
Author :
Takahashi, Norikazu ; Nishi, Tetsuo
Author_Institution :
Dept. of Comput. Sci. & Commun. Eng., Kyushu Univ., Fukuoka, Japan
Abstract :
Convergence property of decomposition methods for solving quadratic optimization problems arising in support vector machines is studied. It is shown that under a mild condition any decomposition method converges to an optimal solution.
Keywords :
convergence of numerical methods; iterative methods; quadratic programming; support vector machines; decomposition methods; global convergence analysis; mild condition; optimal solution; quadratic optimization problems; support vector machines; Computer science; Convergence; Machine learning; Matrix decomposition; Optimization methods; Pattern recognition; Quadratic programming; Signal processing; Signal processing algorithms; Support vector machines;
Conference_Titel :
Circuits and Systems, 2004. ISCAS '04. Proceedings of the 2004 International Symposium on
Print_ISBN :
0-7803-8251-X
DOI :
10.1109/ISCAS.2004.1329911
