Parallel interpolation by using hybrid neural network

In this paper, a basic concept of parallel interpolation technique using neural network architecture and its experimental result are presented. We describe the neural-interpolation method which can execute parallel interpolation by local procedure. A hybrid neural network which is composed of nonlinear units and linear units is used in this study. The proposed method is evaluated by computer simulations for special line interpolation, and it is compared with traditional cubic spline interpolation. As an experimental result, it is shown that the proposed method has a good performance compare to cubic spline. Also, comparative study of parallel interpolation technique by using different network architecture by applying some complicated curved surface interpolation. As a result, the hybrid network architecture has a good performance and accuracy than ordinary sigmoidal one.