Getting order in chaotic cellular neural networks by self-organization with Hebbian adaptation rules

The effects of self-organization using Hebbian adaptation laws were experimentally investigated for two classes of CNN systems. The first class includes autonomous oscillatory neural networks while the second includes chaotic synchronizing CNNs. Allowing interconnection weights to adapt during the state evolution of such systems, higher degree of order is achieved for both classes of systems, and based on this observations we introduce a conjecture regarding the relationship between system entropy and self-organization. This result seems to have a universality characteristic. Within the CNN's framework, the above mentioned phenomena may be exploited as new classes of computational behaviors. Some applications are suggested including image processing, template design, pattern formation and intelligent chaotic synchronization.