Dovbysh A.S., Moskalenko V.V., Rizhova A.S.
Information-extreme method of classification of observation with categorical attributes

An algorithm is proposed for information-extreme machine learning based on the adaptive coding of multitype primary features used in the recognition and optimization of geometric parameters of partitioning the space of secondary (unified) features into equivalence classes in the iterative approximation of the global maximum of an information criterion to its boundary value. © 2016, Springer Science+Business Media New York.

Keywords: categorical attribute, classifier, information criterion, machine learning, pattern recognition, system of control tolerances for recognition features, training matrix, Algorithms, Approximation algorithms, Artificial intelligence, Classifiers, Equivalence classes, Global optimization, Iterative methods, Learning systems, Pattern recognition, Pattern recognition systems, Adaptive coding, Boundary values, Categorical attributes, Extreme machine learning, Extreme method, Information criterion, Iterative approximations, Recognition features, Classification (of information)

Cite:
Dovbysh A.S., Moskalenko V.V., Rizhova A.S. (2016). Information-extreme method of classification of observation with categorical attributes. Cybernetics and Systems Analysis, 52 (2), 56-63. doi: https://doi.org/10.1007/s10559-016-9818-1 http://jnas.nbuv.gov.ua/article/UJRN-0000496947 [In Russian].