一种基于模糊积分的多分类器联合方法

多分类器联合是解决复杂模式识别问题的有效办法。模糊积分是一种多分类器联合算法。但是在用模糊积分对多分类器联合建模时，模糊积分密度函数的选择对系统的性能有着很大的影响。提出一种基于模糊积分和贝叶斯联合的分类器集成方法，该方法利用贝叶斯方法的构造只分密度函数，再利用模糊积分把分类器输出信息联合起来。实验结果表明，该方法比其他方法能够得到更好的识别性能。

Dynamic combination method of multiple classifiers

A combination of multiple classifiers is a powerful solution to the difficult pattern recognition problem. Fuzzy integral is a valid algorithm for combining classifiers. But when modlelling multiple classifers with fuzzy integral, the system's performance is largely based on the fuzzy integral densities. A combination method based on fuzzy integral and Bayes is presented. It gets the fuzzy integral density function through the Bayes method, and combines the output information of all classifers with fuzzy integral. Experiments have shown that this method achieves very promising performance and out-performs other combining approachs.