Support Vector Machine and Its Applications to Function Approximation

Support vector machine is a new machine learning algorith m, based theoretically on statistic learning theory created by Vapnik. Employing the criteria of structural risk minimization, which minimizes the errors betwee n sample data and model data and decreases simultaneously the upper bound of p redict error of model, SVM's generalization is better than others. The character istics of SVM, such as the strong learning capability based on small samples, th e good characteristic of generalization and insensitivity to random noise distur bance, are shown by its applications to function approximation.