Mathematical algorithms for the supervised classification based on fuzzy partial precedence

Authors:
J. Ruiz-Shulcloper;M. Lazo-Cortés
Affiliations:
Instituto de Cibernética, Matemática y Física, E # 309 esq. 15 Habana 10400, Cuba and Centro de Investigatión en Computación-IPN Juan de Dios Batiz s/n esq. Miguel Oth ...;Instituto de Cibernética, Matemática y Física, E # 309 esq. 15 Habana 10400, Cuba
Venue:
Mathematical and Computer Modelling: An International Journal
Year:
1999

Citing 2
Cited 3

Integer-valued problems of transforming the training tables in k-valued code in pattern recognition problems

Pattern Recognition
Determining the feature relevance for non-classically described objects and a new algorithm to compute typical fuzzy testors

Pattern Recognition Letters

Discovering Differences in Patients with Uveitis Through Typical Testors by Class

PKDD '00 Proceedings of the 4th European Conference on Principles of Data Mining and Knowledge Discovery
Two algorithms for architecture pattern recognition

SE '08 Proceedings of the IASTED International Conference on Software Engineering
RGC: A new conceptual clustering algorithm for mixed incomplete data sets

Mathematical and Computer Modelling: An International Journal

Quantified Score

Hi-index	0.98

Visualization

Abstract

Some mathematical algorithms for supervised classification problems are presented in this paper. These algorithms are based on fuzzy partial precedents, and they allow us to work with nonclassically described Objects, i.e., mixed data. These types of descriptions frequently arise in soft sciences. As a rule, the most used methods for solving such classification problems are oriented towards one type of feature, most often quantitative. They do not allow the use of different kinds of features, as a consequence, all variables must be quantitative, or exclusively qualitative. In fact, those methods use, in some way, a distance measure between object descriptions, which follows from the hypothesis of compactness of classes. The proposed models allow the handling of quantitative and qualitative features together, and missing values. They are based on partial evaluation of similarity between objects in a fuzzy environment. They do not use a distance.