A High Order Neural Network Model and Application in Economic Forecast

In this paper, we propose a vector input neural network model. The architecture of this network is composed by two parts: single vector immutiply and mix (de-mix) matrix process. The model can be described as a high dimension neural network operator. Simplify this model bring to a high dimension array as the kernel of the network. The high dimension neural network is usable in many fields especially the input signals are vectors. High dimension immutiply has the physical mean that the victor input channels are filtered at the same sample time. We use this neural network forecasting economic. The economic statistics data such as GDP, CPI composed the input vector and we can use this model (as a high dimension array) to forecast the data of next year or later.