A model of correlative perception and its applications to certain problems of pattern recognition

By correlativity of perception we understand its capability to find out similar configurations of stimuli, and to form high-level configurations with them. It is equivalent to the description of information in terms of generative elements and their transformations. The corresponding mathematical model decomposes certain compound sturctures of data into separate components, simplifying their further identification by a recognition system. The model is applied for problems of automatic notation of performed music – recognition of rhythm, i.e. recurrency in the interchange of sounds under varying tempo, and recognition of chords, i.e. representation of spectrum of a compound musical signal by notes. Possible applications to image processing and simulation of abstract thinking are discussed.