Terminal dynamics approach to cellular neural networks
Valeri Mladenov, Hans Hegt, Arthur H. M. van Roermund · 2002
This paper considers Cellular Neural Networks (CNNs) based on the paradigm of terminal dynamics. They are presented as unpredictable systems, and it is shown that sign strings at the critical points are able to control them. A CNN is able to reproduce a prescribed behavior with a prescribed accuracy by changing a combination of signs, related to interconnections between neighboring neurons. The design approach of CNN template coefficients is considered as well.