Title :
Possibility of synchronization in Hopfield like Chaotic Neural Networks with just adding a single neuron
Author :
Mahdavi, Nariman ; Menhaj, Mohammad B. ; Afshar, Ahmad
Author_Institution :
Electr. Eng. Dept., Amirkabir Univ. of Technol., Tehran, Iran
Abstract :
The ability of chaos appearance in ANN is an effort to simulate the actual behavior of a living cell and it is important to discover effective tuning mechanisms for chaotic and oscillatory activities. In this paper, a new set of sufficient conditions is proposed that guarantees the synchronization of all neuron´s outputs with each other. Furthermore, we consider a possibility of synchronization in all outputs of the Hopfield like Chaotic Neural Networks with just adding a single neuron and adjusting the couplings connected to it. All neurons in this type of network have both self-coupling and non-invertible activation functions which enable them to act as chaotic oscillators. Finally, our claim is evaluated by performing simulations on two illustrative examples.
Keywords :
Hopfield neural nets; chaos; Hopfield like chaotic neural network; artificial neural network; chaotic oscillators; noninvertible activation function; self-coupling activation function; tuning mechanism; Artificial neural networks; Biological neural networks; Chaotic communication; Neurons; Oscillators; Synchronization;
Conference_Titel :
Neural Networks (IJCNN), The 2010 International Joint Conference on
Conference_Location :
Barcelona
Print_ISBN :
978-1-4244-6916-1
DOI :
10.1109/IJCNN.2010.5596986