基于遗传算法优化的神经网络电子邮件信息分类器的研究

结合反垃圾邮件技术的研究,分析了电子邮件数字信息预处理中的特征选择法和将机器学习技术应用于数字信息分类器的方法.针对邮件信息特征向量庞大的问题,提出了GA-CHI特征选择法作为信息预处理,将复杂的邮件数字信息转变成易于机器学习处理的形式.基于BP神经网络电子邮件数字信息分类器,采用遗传算法来优化神经网络邮件数字信息分类器,以进一步提高对中文电子邮件的分类效果.通过对系统的实验分析表明:该文采用的方法能有效地实现对电子邮件数字信息的分类.

E-mail Information Classifier of Neural Network Based on Genetic Algorithm Optimization

Combined with the research on Anti-Spam technology, the feature selection algorithm in pretreatment of e-mail information and the method of applying machine learning technology to digital information classifier is analyzed. In view of the problem that mail message eigenvector is so huge, GACHI feature selection algorithm as pretreatment of information is proposed. It transforms complicated e-mail information into the form which can be easily managed by machine learning. In order to further enhance the effectiveness of Chinese e-mail classification, e-mail information classifier based on BP neural network adopts genetic algorithm to optimize itself. Experimental analysis of the system shows that the method described in the paper can effectively realize the classification of the e-mail information.