Kombinování klasifikátorů pomocí fuzzy metod
In this paper we elaborate a method of combining classifiers by means of the tconorm integral for SVM classifiers ensembles. We compare accuracy of combined classifiers for the case of SVM classifiers and quazi-Sugeno integrals. For t-conorm systems for integration, which we use to determine quazi-S...
Main Author: | |
---|---|
Other Authors: | |
Format: | Dissertation |
Language: | Czech |
Published: |
2007
|
Online Access: | http://www.nusl.cz/ntk/nusl-289162 |
Summary: | In this paper we elaborate a method of combining classifiers by means of the tconorm integral for SVM classifiers ensembles. We compare accuracy of combined classifiers for the case of SVM classifiers and quazi-Sugeno integrals. For t-conorm systems for integration, which we use to determine quazi-Sugeno integrals, we use t-norms from three families of t-norms, Aczél-Alsina family, Dombi family and Frank family. Results of our experiments show that combining classifiers by Sugeno integral is superior to integrals relying on the other considered t-norms. |
---|