The stable learning rates for a two-layer neural network are discussed first by the Lyapunov stability theorem. This two-layer NN can then be incorporated into a fuzzy-neural-network (FNN) for a more efficient tuning process by a new genetic algorithm designed in this paper. The main contribution of this methodology is to reduce the searching time by searching only one learning rate in the FNN. All the equations for tuning both the NN and FNN are fully explained.
|Journal||Proceedings of the IEEE International Conference on Systems, Man and Cybernetics|
|State||Published - 1 Dec 1999|
|Event||1999 IEEE International Conference on Systems, Man, and Cybernetics 'Human Communication and Cybernetics' - Tokyo, Jpn|
Duration: 12 Oct 1999 → 15 Oct 1999