A PC-based software tool for statistical pattern classifier design

Interactive software systems for pattern analysis and recognition represent powerful scientific, research and development tools for pattern recognition system design. In this paper, we discuss one such scientific software system, named PC-PARIS (Pattern Analysis and Recognition Interactive System for PC compatible computers). A variety of pattern recognition and clustering algorithms, implemented in PC-PARIS, is briefly described. As application examples, we present sensitivity analysis of statistical pattern classifiers, implemented in PC-PARIS, to the increasing number of features in the limited training data set conditions, for both synthesized and real data classification.