贝叶斯分类器集成的增量学习方法

针对基于决策树和神经网络的增量学习算法的过量匹配和分类精度有限的缺点,提出了一种基于贝叶斯分类器集成的增量学习方法.综合朴素贝叶斯的增量分类和集成的增量学习方法,采用随机属性选择训练初始SBC(simple Bayesian classifiers),通过判断是否带有类别标签,将增量样本自动分组,并利用遗传算法对结果进行优化.实验结果表明,贝叶斯分类器集成的增量学习方法有效.

Increment Learning Algorithm Based on Bayesian Classifier Integration

An increment learning algorithm based on Bayesian classifier integration is proposed to overcome the shortcomings,overloaded matching and limited classifying precision of the increment learning algorithm based on decision-making tree on a neural network.The increment classifier of simple Bayesian and integrated increment learning algorithm are combined.The SBC(simple Bayesian classifiers) is trained by random property and the increment samples are classified automatically by the tag.The results are optimized by GA(genetic algorithm).The efficiency of the increment learning algorithm based on Bayesian classifier integration has been confirmed by experimentation.