Title :
Genetic Support Vector Classification and Feature Selection
Author :
Mejia-Guevara, I. ; Kuri-Morales, Ángel
Author_Institution :
Inst. de Investig. en Mat. Aplic. y Sist., Univ. Nac. Autonoma de Mexico, Mexico City
Abstract :
An important issue regarding the design of support vector machines (SVMs) is considered in this article, namely, the fine tuning of parameters in SVMs. This problem is tackled by using a self-adaptive genetic algorithm (GA). The same GA is used for feature selection. We validate our results implementing some statistical tests based on single domain benchmark data sets, which are used for comparison with other traditional methods. One of these methods is commonly used for the selection of parameters in SVMs.
Keywords :
genetic algorithms; pattern classification; statistical testing; support vector machines; genetic support vector classification; genetic support vector feature selection; self-adaptive genetic algorithm; statistical test; support vector machines; Artificial intelligence; Benchmark testing; Circuit optimization; Genetic algorithms; Kernel; Machine learning; Static VAr compensators; Statistical analysis; Support vector machine classification; Support vector machines; Self-adaptive Genetic Algorithm; Support Vector Classification;
Conference_Titel :
Artificial Intelligence, 2008. MICAI '08. Seventh Mexican International Conference on
Conference_Location :
Atizapan de Zaragoza
Print_ISBN :
978-0-7695-3441-1
DOI :
10.1109/MICAI.2008.48
