Chaos Control and Neural Classification

Abstract

Abstract Chaotic behaviour in biological neural networks is known from various experiments. The recent finding that it is possible to "control" chaotic systems may help answer the question whether chaos plays an active role in neutral information processing. It is demonstrated that a method for chaos control which was proposed by Pyragas can be used to let a chaotic system act like an autoassociative memory for time signal inputs. Specifically a combined chaotic and chaos control system can reconstruct unstable periodic orbits from incomplete information. The potential relevance of these findings for neural information processing is pointed out.