Please use this identifier to cite or link to this item: http://hdl.handle.net/1959.3/47356

Title

An adaptive recurrent network training algorithm using IIR filter model and Lyapunov theory

Author(s)

Seng, Kah Phooi;  Man, Zhihong;  Wu, Hong Ren;  Tse, K. M.

Abstract

A new approach for the adaptive algorithm of a fully connected recurrent neural network (RNN) based upon the digital filter theory is proposed. Each recurrent neuron is modeled by using an infinite impulse response (IIR) filter. The weights of each layer in the RNN are updated adaptively so that the error between the desired output and the RNN output can converge to zero asymptotically. The proposed optimization method is based on the Lyapunov theory-based adaptive filtering (LAP) method [9], The merit of this adaptive algorithm can avoid computation of the dynamic derivatives that is rather complicated in the RNN. The design is independent of the stochastic properties of the input disturbances and the stability is guaranteed by the Lyapunov stability theory. Simulation example of the nonstationary time series prediction problem is performed. The simulation results have validated the fast tracking property of the proposed method.

Publication type

Book chapter

Source

Recent advances in computers, computing and communications / Nikos Mastorakis and Valeri Mladenov (eds.), pp. 287-289

Publication year

2002

Keyword(s)

Adaptive algorithms;  Adaptive filtering;  Backpropagation;  Computational methods;  Computer simulation;  IIR filters;  Infinite impulse response filters;  Learning algorithms;  Lyapunov methods;  Lyapunov stability theory;  Optimization;  Real time systems;  Real-time recurrent learning;  Recurrent neural networks;  RNNs;  RTRL;  Stochastic properties

Publisher

World Scientific and Engineering Academy and Society

ISBN

9608052629

Peer reviewed