Full Article: PDF
Scientific Object Identifier: http://s-o-i.org/1.1/TAS-01-57-6
DOI: https://dx.doi.org/10.15863/TAS.2018.01.57.6
Language: English
Citation: Kamilov MM, Hudayberdiev MK (2018) FORMATION OF A QUALITATIVE DESCRIPTION OF THE TRAINING SET IN SOLVING THE RECOGNITION PROBLEM. ISJ Theoretical & Applied Science, 01 (57): 33-37. Soi: http://s-o-i.org/1.1/TAS-01-57-6 Doi: https://dx.doi.org/10.15863/TAS.2018.01.57.6 |
Pages: 33-37
Published: 30.01.2018
Abstract: The paper discusses the possibilities of improving the quality of the recognition algorithm based on partial precedent, by the original pre-training procedures. The peculiarity of this algorithm is that as precedents only such "anchor points" of a pattern that ensuring the following conditions are left: the distance from any point on the training set of i-th pattern to their nearest precedent is less than the distance to the nearest precedent of another pattern. This set of precedents provides unmistakable recognition of all samples of the training set. Thus, the probability of correctly separating of classes increases significantly. The set of dedicated training samples gives a chance to improve the level of reliability of data mining. One of species of tulip has been chosen as object of research. This process is carried out via morphological features of tulip. The information about tulip is obtained from Central herbarium of institute of Botany of the Uzbek Academy of sciences.
Key words: data mining, pattern recognition, algorithm of partial precedents, precedent, training set, etalon objects, classification, clustering.
|
