Forecasting Exchange Rate Based on Chaos Particle Swarm Optimization——Wavelet Support Vector Machine

Nowadays,the common parameter optimization methods of support vector machine(SVM) are easy to lapse into local extremum,and the approximation accuracy of its frequently-used kernel functions also needs to be improved.Based on the ergodicity and stochastic property of chaos mapping as well as the local analysis and feature extraction abilities of wavelet transform,an algorithm is presented which is named as chaos particle swarm optimization wavelet SVM(CPSO-WSVM).This algorithm is used to construct exchange rate forecasting model.The experimental results show that CPSO-WSVM method has good application effect,which obtains much higher forecasting precision and efficiency than the traditional particle swarm optimization-Gaussian kernel SVM(PSO-GSVM).