抄録
This paper studies how the generalization ability of modeling of the dynamic systems can be improved by taking advantages of second order derivatives of the criterion function with respect to the external inputs that can be considered as a direct implementation of the well-known regularization technique. Computation of second order derivatives of Universal Learning Network for a dynamic network are discussed. Simulation studies of a nonlinear dynamic system and a real robot system are carried out to demonstrate the effectiveness of the proposed method. Simulation results show that the proposed method can improve the generalization ability of neural networks sufficiently by selecting an appropriate regularization parameter.
