A Consideration on Application of Support Vector Machine for Non-linear Optimization

Abstract

In this Paper, we consider the method of applying Support Vector Machine, which is attracting attention as the nonlinear discriminant analysis, to optimize model parameters of a complex system simulator for civil planning by proposing the simple algorithm and trial calculation of test-functions. As the conclusion, when the case of the time for function approximation is shorter than the probability optimization techniques represented by simple GA, the possibility of shortening calculation time was shown by using the response surface methodology and Support Vector Machine simultaneously.