System identifications by SIRMs models with linear transformation of input variables

Abstract

This paper shows the effectiveness of the model proposed in the previous paper for system identifications. In the first simulation, which is for EX-OR, the fundamental idea of the proposed model is explained. In the second simulation, which is for classification problems for dataset of Iris, Wine, Sonar and BCW known as benchmark problems, the capability of the model is evaluated for involving a large number of input variables. In the third simulation, as one of control problems, numerical simulation for obstacle avoidance problem is performed. In these simulations, it is shown that the proposed model outper forms conventional models interms of system identifications.