An improved approach based on least squares support vector machine（LS-SVM） is proposed for solving nonlinear ordinary differential equations initial value. LS-SVM model is improved by using the differentiable radial basis function（RBF） kernel function and transform the model with derivative form of kernel into an optimization problem to solve. The proposed approach can obtain the optimal representation of the solution in the primal-dual setting and provides a continuous-differential form approximate solution with high precision. In addition

numerical experiments are presented and compared with exact solutions to confirm the validity and accuracy of the proposed approach.