论文部分内容阅读
利用K均值聚类和增量学习算法扩大训练样本规模,提出一种改进的mRMR SBC.一方面,利用K均值聚类预测测试样本的类标签,将已标记的测试样本添加到训练集中,并在属性选择过程中引入一个调节因子以降低K均值聚类误标记带来的风险.另一方面,从测试样本集中选择有助于提高当前分类器精度的实例,把它加入到训练集中,来增量地修正贝叶斯分类器的参数.实验结果表明,与mRMR SBC相比,所提方法具有较好的分类效果,适于解决高维且含有较少类标签的数据集分类问题.
Using K-means clustering and incremental learning algorithm to expand the training sample size, an improved mRMR SBC is proposed.On the one hand, K-means clustering is used to predict the class label of the test sample, the marked test sample is added to the training set, and In the process of attribute selection, an adjustment factor is introduced to reduce the risk of false labeling of K-means clustering.On the other hand, selecting from the test sample set an example that can improve the accuracy of the current classifier and adding it to the training set The parameters of Bayesian classifier are modified incrementally.The experimental results show that the proposed method has better classification results than mRMR SBC and is suitable for solving data classification problems with high dimension and fewer types of labels.