Two adaptation methods of artificial neural networks
Baocheng Sun, Zhifang Zhang · 2002
In order to cope with the existing errors in modeling of multilayered feedforward neural networks (MLF), this paper presents two adaptation methods of artificial neural networks: feedback adaptation and Taylor series expansion based adaptation, based on the trained MLF with some modeling errors. Simulation results show that the proposed two adaptation methods give good error-reduction in modeling and forecasting of MLF.>