Parameters selection of SVM for function approximation based on Differential Evolution

DOI

10.2991/iske.2007.90How to use a DOI?

Keywords

machine learning, support vector machines, artificial neural networks, differential evolution, function approximation

Abstract

Support vector machines (SVM) is a new machine learning method, and it has the ability to approximate nonlinear functions with arbitrary accuracy. Right setting parameters are very crucial to learning results and generalization ability of SVM. In this paper, parameters selection is regarded as a compound optimization problem and a modified differential evolution (MDE) algorithm is applied to search the optimal parameters. The modified differential evolution adopts a time-varying crossover probability strategy, which can improve the global convergence ability and robustness of the algorithm. Various examples are simulated and the experiment results demonstrate that this proposed approach has better approximation performance than other approaches

Copyright

© 2007, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - ZHOU Shaowu
AU  - WU Lianghong
AU  - YUAN Xiaofang
AU  - TAN Wen
PY  - 2007/10
DA  - 2007/10
TI  - Parameters selection of SVM for function approximation based on Differential Evolution
BT  - Proceedings of the 2007 International Conference on Intelligent Systems and Knowledge Engineering (ISKE 2007)
PB  - Atlantis Press
SP  - 529
EP  - 535
SN  - 1951-6851
UR  - https://doi.org/10.2991/iske.2007.90
DO  - 10.2991/iske.2007.90
ID  - Shaowu2007/10
ER  -