Some characteristics of fuzzy integrals as a multiple classifiers fusion method

Authors:
Huimin Feng;Xuefei Li;Tiegang Fan;Yanju Chen
Affiliations:
Department of Mathematics and Computer Science, Hebei University, Baoding, Hebei, P.R. China;College of Science, Hebei Agriculture University, Baoding, Hebei, P.R. China;Department of Mathematics and Computer Science, Hebei University, Baoding, Hebei, P.R. China;Department of Mathematics and Computer Science, Hebei University, Baoding, Hebei, P.R. China
Venue:
ICMLC'05 Proceedings of the 4th international conference on Advances in Machine Learning and Cybernetics
Year:
2005

Citing 8
Cited 0

Optimal combinations of pattern classifiers

Pattern Recognition Letters
Some aspects of Dempster-Shafer evidence theory for classification of multi-modality medical images taking partial volume effect into account

Pattern Recognition Letters
Optimal linear combinations of neural networks

Neural Networks
Using neural networks to determine Sugeno measures by statistics

Neural Networks
Soft combination of neural classifiers: a comparative study

Pattern Recognition Letters
A genetic algorithm for determining nonadditive set functions in information fusion

Fuzzy Sets and Systems - Special issue on fuzzy measures and integrals
Fusing Neural Networks Through Space Partitioning and Fuzzy Integration

Neural Processing Letters
Multiple network fusion using fuzzy logic

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

Fuzzy integrals have attracted the attention of many researchers as a solution for expressing the interactions between classifiers in multiple-classifier fusion. In a classifier fusion system based on fuzzy integrals, the fuzzy measures will have a major impact on a system’s performance. Much work has been carried out by numerous authors on how to determine the fuzzy measures to improve results. Our paper presents some new characteristics of multiple-classifier fusion based on fuzzy integrals. This paper discusses the conditions under which the fusion system must give the incorrect classification and that the fusion system can give the correct classification even if all classifiers have given an incorrect classification. It will be helpful for improving classifier fusion systems and designing classifiers in application.