摘要
为提高分类器识别率,减少标注样本使用数量,提出一种基于朴素贝叶斯的半监督学习方法。研究基于该方法的分类器分类效果,采用遥感影像数据作为训练和测试集,与基于朴素贝叶斯的全监督学习分类器分类效果作比较。实验结果表明,当标注样本与非标注样本比例在1:2~1:9时,半监督学习可以利用比全监督学习更少的标注样本,达到更高的分类精度。
This paper proposes a method of probabilistic semi-supervised learning in order to improve accuracy of classifier and reduce cost of labeled sample.Research is based on the method,and classification results are compared with that based on naive Bayesian full-supervised learning.This paper uses remote sensing image data as the training and test sample set.Experimental results show that when the proportion of labeled sample to unlabeled sample is between 1:2 and 1:9,classification based on semi-supervised learning can get higher accuracy than that based on full-supervised learning using much less labeled sample.
出处
《计算机工程》
CAS
CSCD
北大核心
2010年第20期167-169,共3页
Computer Engineering
基金
国家"863"计划基金资助项目(2006AA12Z105)
国家"973"计划基金资助项目(2006CB701303)。
关键词
朴素贝叶斯
半监督学习
遥感影像分类
naive Bayesian
semi-supervised learning
remote sensing image classification