【摘要】 针对标准遗传算法的不稳定性、准确性低等问题,为了提高遗传分类算法的稳定性和准确性,基于贝叶斯算法的有关理论,提出一种新的遗传算法分类方法。将初始样本集随机的分成数量相等的几组,通过朴素贝叶斯算法从初始样本集中选出部分"区分度"比较高的样本作为新的样本集,通过改进的遗传算法对选出的新样本集进行处理,从而得到最优分类规则。通过两种算法的组合对数据分类时,使分类的稳定性和准确性得到了明显的改善。仿真实验结果表明,该算法有较高的稳定性和准确性。更多还原

【Abstract】 Aimed at the problems of instability and low accuracy in standard genetic algorithm,in order to improve the stability and accuracy of the genetic classification algorithm,based on theory of the Bayesian algorithm,a new method of genetic algorithm classification is presented.First,the initial sample set is divided into randomly groups of equal number.Second,select some samples of which the "discrimination" is relatively high from the initial sample set by the naive Bayesian algorithm as a new sample set.Third,the new sample set through the improved genetic algorithm is processed to get the optimal rule.Through the combination of two algorithms for data classification,the stability and accuracy of the classification are improved obviously.The result of simulation indicates that this algorithm has higher stability and accuracy.更多还原