MLPs Vs. SVM, QP problem，geometrical interpretation， primal space，dual space，fi-function，slack variance（L1，L2；origal,LS SVM），linear kernel，RBF kernel（Gaussian），VC dimension，LS SVM :Fisher, sparse achieving, GP,ARD,RKHS,weighted LS SVM, 3 Levels of ibayesian nference, Nystrom, fixed size, combination, pca,cca,recurrent, optimal control.