A Comparation Of Common Fuzzy Integrals In Multi-classifier Fusion

The accuracy and robustness of classification system can be improved through fusion of multiple classifiers. Fuzzy Integral, which can express the importance of individual classifiers and the interaction among classifiers, is widely accepted and applied as an aggregation tool in multi-classifier fusion. There are two keys to determine. One is the type of fuzzy integral. There are four types of fuzzy integral to use in fusion, Sugeno integral, Choquet integral, upper integral and lower integral. Another is the fuzzy measure. In multi-classifier fusion based on fuzzy integrals, fuzzy measures have much influence on the performance of fusion systems. If the fuzzy measures are well defined, the accuracy of classification can be improved distinctly. However, if the fuzzy measures are badly defined, it is possible that most classifier's accuracy of classification is higher than the fusion system's accuracy. For the fuzzy integral, the type of fuzzy integral is usually specified in advance. Choquet integral is often the choice. There is no conclusion which type integral is more suitable for classifier fusion. About the fuzzy measure, many researchers have done much work and proposed many types of fuzzy measure and methods to determine fuzzy measures. This thesis is to compare four fuzzy integrals in the multiple-classifier fusion from the definition, relationship and as operators. We hope it can give some helpful conclusion. According the theoretical and experimental analysis, this thesis gives the conclusion that the proper order for multiple-classifier fusion is Choquet integral, lower integral, the upper integral and Sugeno integral.