Artificial intelligence

Scientific journal

ISSN 2710-1673

ONLINE: ISSN 2710-1681

Select your language


Тестова база даних і методика порівняльної оцінки класифікаторів

Tatur M.1, Bayrak S.2, Ostrovsky V.2
1 Belorussian State Univercity of Informatics and Radioelectronics
2 Belarusian State University of Informatics and Radioelectronics BSUIR of Belarus

Full text (PDF)

UDC: 381.3
Publication Language: Russian
Stuc. intelekt. 2008; 13(3):501-504

Abstract: The problem of proper comparative classifier rating was researched. Authenticity was taken as efficiency criteria for binary-output classifiers and accuracy – for fuzzy classifiers. Several specific classification signs were highlighted, which resulted in definition of 8 typical tasks. For each of these tasks function value could be calculated both for random and determined input data. Experiment set system for classifiers was proposed.

Keywords:

References:

View full text (PDF)