Zi-Shen Pill Quality Assessment Based on Rough Sets and Back-Propagation Neural Network

To discriminate the quality on traditional Chinese medicines Zi-shen pill real-time, according to the characters of Zi-shen pill finger printer, the basic concepts of rough set are introduced briefly. For rough sets can only deal with discrete data, the discretization of data is the key factor in the rough sets applied in quality assessment, we present a method of discretization based on cluster category which combined with the characteristic of rough sets and back-propagation neural network, its generalization is well. Using methods rough sets and artificial neutral network to assess Zi-shen pill without any additional prior model assumption, rough sets data analysis can eliminate the redundancy of attributes and its value, identify the dependence in the attributes. We get a collective production rules about the chemical pattern classification system from sample data. When the model of chemical pattern classification is built by these rules, its meaning is very understandable in chemical domain, and the prediction of the model is also well.