Application Of Fuzzy Integral In Multi-classification Fusion

Fuzzy Integral is an aggregation tool in multi-classifier fusion, which can improve the accuracy of classification and the robustness of systems. In the model of multi-classifier fusion based on fuzzy integrals respect to lambda-fuzzy measures, the fusion results are heavily dependent on fuzzy densities that represent the importance of individual classifier to the fusion results. Some methods only take the account of accuracy in determining of fuzzy densities, but other uncertainty which can also reflects the performance of a classifier are not concerned. In this thesis, two types of uncertainties of classifiers are analyzed: accuracy, and the distinguish ability of classification. The property of distinguish ability is discussed in detail. The similarity is introduced and used to measure it. Finally, a new methodology for determining fuzzy densities is proposed through integrating accuracy, the distinguish ability of the classifiers. This methodology is verified to be practical by experiments.