����DE�Ħ�-SVRM�����Ż��о�

ժҪ
ͼ/��
�ο��
�� (15)

ȫ��: PDF (421 KB) HTML (0 KB)
��: BibTeX | EndNote (RIS)

ժҪ ��Ż��Ǧ�-֧��ع��о��Ҫ��⣬�䱾��һ��Ż��Ĺ�̡��ڲ��ݻ��㷨��Ż��ϵ��Ч�ԣ��Բ��ݻ��㷨Ѱ�ż��ɵĦŪ�֧��ع��Ż��㷨Ӧ��Ӱ��ı�׼��Ŵ��㷨��Ⱥ�㷨��Ⱥ�㷨��֧��Ż��ķ��ʵ��Աȱ��DE�㷨��ȷ��ĦŪ�֧��ع��нϺõ�Ԥ��ܡ�

	��

	�ѱ��Ƽ��
	��ҵ��
	��ù��
	E-mail Alert
	RSS
	��
	��
	�Ի�

�ؼ�� �� -֧��ع��, ��Ż�, ��ݻ�

Abstract��Parameters optimization is an important issue in the research of �Ū�SVRM, and it nature is an optimization selection process. Motivated by the effectiveness of Differential Evolution algorithm on optimization problem, a new automatic searching methodology based on Differential Evolution (DE) algorithm was proposed. Experimental results assess the feasibility of the proposed approach for data sets with several noises and compared with SVRM model optimization method based on Genetic Algorithm(GA), Ant Colony Optimization(ACO) and Particle Swarm Optimization(POS) demonstrate improvement of generalization performance.

Key words�� -SVRM Parameter Optimization DE

�ո��: 2008-02-28

ͨѶ��: �� E-mail: alphah@qq.com;nkalpha@126.com

��ñ��:

�� Ի�. ��DE�Ħ�-SVRM��Ż��о�[J]. ��Ӧ��, 2008, 28(8): 2074-2076. Wang Jin-Lin Hui Zhao. Parameters optimization of ��-support vector regression machine based on differential evolution algorithm. Journal of Computer Applications, 2008, 28(8): 2074-2076.

��ӱ��:

http://www.joca.cn/CN/ �� http://www.joca.cn/CN/Y2008/V28/I8/2074