Abstract:

The method of a support vector regression machine was  developed for solving single output regression problems. However many questions belong to multioutput regression in practice. In order to deal with these  problems, the support vector regression machine was extended to multioutput by parallel operation. And it was applied to solve the problem of timedependent variational inequalities which describe dynamic traffic equilibrium. The result of one experiment showed that it had faster calculation speed and better fitting effect than a singleoutput support vector regression machine and linear interpolation. The algorithm promoted the study of a multioutput support vector regression machine and provided a novel means to solve the problem of timedependent variational inequalities.
 

Key words: linear programming support vector regression machine, multi-output support vector regression machine(M-SVR), timedependent variational inequality, dynamic traffic equilibrium problem