DETERMINISTIC LEARNING AND RAPID DYNAMICAL PATTERN RECOGNITION

Abstract

Recognition of temporal/dynamical patterns is among the most difficult pattern recognition tasks. In this paper, based on a recent result on deterministic learning theory, a unified, deterministic approach is proposed for effective representation and rapid recognition of dynamical patterns. Firstly, it is shown that time-varying dynamical patterns can be effectively represented in a time-invariant and spatially-distributed manner through deterministic learning. Then, by characterizing the similarity of dynamical patterns based on the system dynamics inherently within them, a dynamical recognition mechanism is proposed. Rapid recognition of dynamical patterns can be implemented when state synchronization is achieved according to a kind of indirect and dynamical matching on system dynamics. The synchronization errors can be taken as the measure of similarity between the test and training patterns. The significance of the paper is that the problem of dynamical pattern recognition is turned into a problem of stability and convergence of a closed-loop recognition system, so that a completely dynamical approach is presented for rapid recognition of dynamical patterns.