基于关联规则的激光数据主流特征分类

Classification of mainstream features of laser data based on association rules

为了解决传统基于神经网络方法的激光数据主流特征分类过程存在分类失衡以及精度差的问题,提出基于关联规则的激光数据主流特征分类方法,其通过主流特征抽样法提升少数类数据中与类关联性强的数据样本的数量,降低多数类数据中与类关联性弱的数据样本的数量,使得激光数据主流特征得到加强,确保数据类间保持平衡。采用基于趋势度的动态关联规划挖掘算法,获取激光数据主流特征间的动态关联规则,基于该关联规则塑造SVM分类器,实现激光数据主流特征的分类。实验结果说明,所提方法具有较高的分类精度和平稳性,分类质量教优。

In order to solve the laser data of neural network method exists imbalance and inaccuracy of the classification based on the characteristics of the mainstream laser,data classification method was proposed based on association rules,the main feature of sampling method to enhance the number of strong data samples of the minority class data,reduce the number associated with the class of weak data samples of the majority class data,the mainstream characteristic of laser data has been strengthened,keeping the balance between data types.The dynamic association planning mining algorithm based on trend degree is used to obtain the dynamic association rules between the main features of laser data.The SVM classifier is constructed based on the association rules,and the main features of laser data are classified.Experimental results show that the proposed method has higher classification accuracy and stability,and better classification quality.