基于极速学习的Choquet模糊积分分类器

Choquet fuzzy integral classifier based on extreme learning machine

在交互环境下,模糊积分分类器具有良好的分类性能.如何确定在属性集幂集上定义的模糊测度是模糊积分分类器中的一个关键问题.当属性的个数增加时,计算复杂度呈指数级增长.为了解决这一问题,借鉴极速学习机算法中权重向量随机确定的思想,提出了ELM-Choquet模糊积分分类器.实验结果表明,和Choquet模糊积分分类器相比,该算法具有较优的分类性能.

In the interactive environments,fuzzy integral classifier has shown its good classification performance. A key problem in the fuzzy integral classifier is to determine the fuzzy measure defined in the power set of attributes. The computational complexity increases exponentially as the number of attributes increases. In order to solve the problem, drawing lessons from the idea of random determination of weight vector in extrem learning machine algorithm, this paper puts forward an ELM-Choquet fuzzy integral classifier. The experiment result shows that compared with the Choquet fuzzy integral classifier, this algorithm has better classification performance.