A model for selection of attributes for automatic pattern recognition. Stepwise data compression monitored by visual classification

摘要

Based on a pattern recognition study of human chromosomes a model is proposed for selection of attributes for automatic pattern recognition. The basic concept is a stepwise data compression, each step being monitored for loss of significant information by visual classification of the compressed pattern. The attributes resulting from a number of data compression steps will retain the information allowing the patterns to be classified automatically.