基于神经网络和粗糙集规则的提取方法

在利用粗糙集对连续性数据进行分类规则挖掘时，需要对数据进行离散化处理，但是离散结果往往会破坏原有数据的隐含信息，提取的分类规则质量难以保证。该文设计了一种基于自组织人工神经网络与粗糙集理论的分类规则提取方法，利用神经网络自动分类的功能，对离散前后的数据进行分类，比较两次分类结果是否一致，当达到一致性结果后，再利用粗糙集理论对数据约简，进行规则提取，有效地解决了原始数据信息丢失的问题，通过实例证明了该方法的合理性。

Extracting Rules Based on Artificial Neural Networks and Rough Sets Theory

The continuous value is discretized before using the rough set method to mine the classification rules. But more information concealed in the original data is lost after the discretization, the quality of the extracted classification rules is very poor. A new method based on self-organizing artificial neural networks and rough set theory is designed to extract classification rules of continuous value. Because self-organizing artificial neural networks can train themselves and make an auto-classification on the input mode, it is used twice to classify the data before and after discretization. It extracts the rules by rough sets reduction until the results of two classifications are consistent. Through the analysis of case studies, the rationality of the extraction rule is testified.