Stability, Optical Flow and Stochastic Resonance in Cellular Wave Computing
Levente Török, B. Eng · 2005
Just as many of us, I am also prone to handle engineering achievements as being a construction built from solid, reliable, brick-like components. Naturally, this not true in all senses. However, just like living species of mother Nature, electrical systems may not require perfect predictability in all aspects to operate correctly but only in the important ones. Hence, the central questions are ”what is important?” and ”how to design systems for it?”. In this aspect, the stability of the systems we use, is a must ; the applicability is a need ; the noise tolerance is a desire. In the light of these findings, I concentrated on these questions in my work with special regard to a nonconventional computational paradigm, named Cellular Neural/Non-linear Networks Universal-Machine (CNN-UM) paradigm. First, in response to the question how we can use this constuction, I dealt with questions of stability of a newly invented version of the CNN-UM. Then, answering the quesion what it can be used it for, I applied the CNN to calculate a computationally heavy task, by the application of which it was accellareted by several orders of magnitudes. Finally, I dealt with the question whether it can be made noise robust, and I found that the system can even produce the so called stochastic resonance phenomena in which it can be optimized against noise by the assistance of the noise itself.