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

作     者：Chen, CLP Wan, JZ 

作者机构：Wright State Univ Dept Comp Sci & Engn Dayton OH 45435 USA USAF Wright Lab Mat Directorate MLIM Wright Patterson AFB OH 45433 USA Lexis Nexis Data Cent Dayton OH 45343 USA 

出 版 物：《IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS PART B-CYBERNETICS》 (IEEE Trans Syst Man Cybern Part B Cybern)

年 卷 期：1999年第29卷第1期

页      面：62-72页

核心收录：

基　　金：Air Force, (F33610-D-5964, N00014-92-J-4096) Office of Naval Research, ONR, (F49620-94-0277) Office of Naval Research, ONR Air Force Office of Scientific Research, AFOSR 

主　　题：Heuristic algorithms Neural networks Radial basis function networks Backpropagation algorithms Function approximation Linear systems Laser modes Chaos Nonlinear systems Equations 

摘      要：A fast learning algorithm is proposed to find an optimal weights of the flat neural networks (especially, the functional-link network). Although the hat networks are used for nonlinear function approximation, they can be formulated as linear systems. Thus, the weights of the networks can be solved easily 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 added enhancement node. A dynamic stepwise updating algorithm is proposed to update the weights of the system on-the-fly. The model is tested on several time-series data including an infrared laser data set, a chaotic time-series, a monthly flour price data set, and a nonlinear system identification problem. The simulation results are compared to existing models in which more complex architectures and more costly training are needed. The results indicate that the proposed model is very attractive to real-time processes.