Title :
Supervised adaptive resonance theory neural network for modelling dynamic systems
Author :
Pham, D.T. ; Sukkar, M.F.
Author_Institution :
Sch. of Eng., Univ. of Wales Coll. of Cardiff, UK
Abstract :
A supervised neural network, SMART2, has been developed which can be used with the ART2 algorithm for modelling discrete dynamic systems. A new layer has been added as a higher transformation stage to provide an output mapping field. The connection between the new field and the category field has been made by long term memory adaptive filters. Top-down adaptive filters in the new field have been employed to code the output prototype. Error equations have been derived to trace errors in the model and train the new network. The proposed network has been shown in simulation to be able to represent arbitrary dynamic systems. Results presented in this paper demonstrate the effectiveness of the network
Keywords :
ART neural nets; adaptive filters; content-addressable storage; discrete time systems; learning (artificial intelligence); modelling; SMART2; adaptive resonance theory; discrete dynamic systems; dynamic system modelling; long term memory; output mapping field; supervised neural network; top-down adaptive filters; Adaptive filters; Adaptive systems; Data structures; Equations; Feedback; Neural networks; Prototypes; Resonance; Subspace constraints; Supervised learning;
Conference_Titel :
Systems, Man and Cybernetics, 1995. Intelligent Systems for the 21st Century., IEEE International Conference on
Conference_Location :
Vancouver, BC
Print_ISBN :
0-7803-2559-1
DOI :
10.1109/ICSMC.1995.538157
