论文部分内容阅读
粗糙集和神经网络作为不确定性计算的两种重要工具,它们具有很强的互补性。在分析了两种理论的特点之后,得出了一种多神经网络分类器的组合方法,新方法根据对数据集进行约简的结果得到多个与数据相关的且相互独立的神经网络分类器,然后根据属性重要性概念将多个分类器组合起来。对比实验证明,该分类器具有较好的分类效果和性能。
Rough set and neural network are two important tools for calculating uncertainty. They have strong complementarity. After analyzing the characteristics of the two theories, a combination method of multi-neural network classifier is proposed. The new method obtains a number of data-dependent and independent neural network classifiers based on the result of the data set reduction , And then combine multiple classifiers according to the concept of attribute importance. Comparative experiments show that the classifier has better classification performance and performance.