Study on Optimization of SVR Parameters Selection Based on PSO

The regression accuracy and generalization performance of the support vector regression (SVR) models depend on a proper setting of its parameters. An optimal selection approach of SVR parameters was put forward based on particle swarm optimization (PSO) algorithm. Furthermore, a comparison was made between the performance of PSO parameter selection and cross validation (CV) and leave-one-out (LOO) method on various data sets, such as a sin c function with additive noise and a SVR model of the product concentration in fermentation process. Simulation results show that the optimal selection approach based on PSO is available and the PSO-SVR model has superior learning accuracy and generalization performance.