基于遗传算法的朴素贝叶斯分类

朴素贝叶斯分类器是一种简单而高效的分类器,但是其属性独立性假设限制了对实际数据的应用。提出一种新的算法,该算法为避免数据预处理时,训练集的噪声及数据规模使属性约简的效果不太理想,并进而影响分类效果,在训练集上通过随机属性选取生成若干属性子集,并以这些子集构建相应的贝叶斯分类器,进而采用遗传算法进行优选。实验表明,与传统的朴素贝叶斯方法相比,该方法具有更好的分类精度。

Naive Bayes Classification Based on Genetic Algorithms

Naive Bayes classifier is a simple and effective classification method,but its attribute independence assumption makes it unable to express the dependence among attributes in the real world.To avoid the direct influence of feature reduction from data pre-processing on the performance of classification,a new algorithm is introduced in this paper.It makes use of random feature selection to generate several feature subsets from the whole training set,and constructs Bayesian classifiers with the feature subsets,and then optimizes the Bayesian classifiers by using genetic algorithms.Compared with the traditional Naive Bayes methods,the algorithm has better classification precision.