Title :
A method of parameters selection with higher accuracy for SVM
Author :
Fangbin Wang ; Dahua Li
Author_Institution :
Dept. of Mech. & Electr. Eng., Anhui Univ. of Archit., Hefei, China
Abstract :
Based the analysis of the effect of kernel parameters and penalty parameters on the performance of support vector machine(SVM), the paper has proposed a new method of hydroid simulated annealing technology. The experiment run on the datasets of UCI with the algorithm has shown the result with higher accuracy.
Keywords :
pattern classification; simulated annealing; support vector machines; SVM; UCI datasets; classification method; hydroid simulated annealing technology; kernel parameters; parameters selection; penalty parameters; support vector machine; Accuracy; Classification algorithms; Kernel; Optimization; Pattern recognition; Signal processing algorithms; Support vector machines;
Conference_Titel :
Advanced Computational Intelligence (ICACI), 2012 IEEE Fifth International Conference on
Conference_Location :
Nanjing
Print_ISBN :
978-1-4673-1743-6
DOI :
10.1109/ICACI.2012.6463161
