Acceleration of learning in hybrid neural networks: A novel approach for the design of brain chaosmakers
Berj L. Bardakjian, A.W.L. Chiu · 2005
Epileptic seizures correspond to episodes of increased rhythmicity of the normally chaotic activity in biological neural networks. We propose to use hybrid neural networks where artificial neural networks are used to control the biological neural networks by learning their different states. The learning is dramatically accelerated when using a conjugate gradient method in conjunction with the Fletcher-Reeves method of optimization.