Volume 17, No. 1, 2020

An Algorithm for Classification, Localization and Selection of Informative Features in the Space of Politypic Data


Akhram Khasanovich Nishanov, Bakhtiyorjon Bakirovich Akbaraliev, Samandarov Batirbek Satimovich, Akhmedov Oybek Kamarbekovich and Tajibaev Shukhrat Khudaybergenovich

Abstract

Dimensionality reduction and feature subset selection are very important and challenging issues in preliminary processing of the large amount of data for its intellectual analysis, pattern recognition and clustering. In particular, the relevance of these issues will only grow if the preliminary data is derived from real-life and defined by qualitative indictors. Similarly, when problem is related to the selection of complex of important features for classification and localization of agricultural crops, the results are immediately reflected in practice. Thus, the problem would appear unclear should the initial data be polytypic i.e. defined by quality indicators and quantitative features. To address the problem, algorithms and programs have been developed for determining the degree of similarity of the objects based on analysis of the existing literature, and then textual, nominal and quantitative definition of the features in the form of qualitative indicators, and mathematical interpretation of the problem.


Pages: 341-364

DOI: 10.14704/WEB/V17I1/WEB17009

Keywords: Classification and Localization of Objects, Qualitative and Quantitative Features, Evaluation of Objects and Features, Feature Selection

Full Text