Case- and rule-based algorithms for the contextual pattern recognition problem

Abstract

This paper deals with the concept of information unification and its application to the contextual pattern recognition task. The concept of the recognition and the rule-based algorithm with learning, based on the probabilistic model is presented. The machine learning algorithm based on statistical tests for the recognition of controlled Markov chains is shown. Idea of information unification via transforming the expert rules into the learning set is derived. © Springer-Verlag Berlin Heidelberg 2003.