QPSO算法在朴素贝叶斯分类上的应用研究

朴素贝叶斯分类假定类条件独立,使得所选数据集的条件属性集在预处理时必须进行属性约简,如果处理不当,就会造成分类的不准确.本文分别对在训练集上随机选取的属性子集组成粒子,构造适应度函数,从而构建了朴素贝叶斯分类器,并利用量子粒子群算法对分类效果进行择优操作.实验证明,其分类效果优于传统的朴素贝叶斯分类方法.

An Application Study of QPSO Algorithm on Naive Bayesian Classification

The Bayesian classification approach assume that its attribute is independent,so the feature reduction of condition attribute set must be handled in data preprocess.In this process,data classification will not be accurate ifit can not be handled properly.In this paper,random selected attributes of the training sets are used to form particles and a fitness function is set up at the same time,and therefore the naive Bayesian classifiers is constructed.At the same time,QPSO algorithm is used to optimize the effect of classifying.Experiment shows that this method has better performance than that of traditional naive Bayesian method.