A rapid learning and dynamic stepwise updating algorithm for flat neural networks and the application to time-series prediction

A fast learning algorithm is proposed to find an optimal weights of the fla t neural networks (especially, the functional-link network). Although the h at networks are used for nonlinear function approximation, they can be form ulated as linear systems. Thus, the weights of the networks can be solved e asily using a linear least-square method. This formulation makes it easier to update the weights instantly for both a new added pattern and a new adde d enhancement node. A dynamic stepwise updating algorithm is proposed to up date the weights of the system on-the-fly. The model is tested on several t ime-series data including an infrared laser data set, a chaotic time-series , a monthly flour price data set, and a nonlinear system identification pro blem. The simulation results are compared to existing models in which more complex architectures and more costly training are needed. The results indi cate that the proposed model is very attractive to real-time processes.