Programmable functional schemes for the recognition of ordered objects

摘要

Algorithms of pattern recognition that use relations of partial and linear order on a set of objects are considered. Decision rules are formulated using minimal subsets of the training sample that are sufficient for the sample restoration with the use of order relations. Methods of constructing decision rules and their descrip- tion in terms of k-valued functions on a basis of elementary functions are studied, together with methods of their realization by automatic programmable schemes in languages of functional programming.