基于分层高斯混合模型的半监督学习算法.pdf

第 41 卷 第 1 期 计 算 机 研 究 与 发 展 Vol 41 ,No 1 2004 年 1 月 J OURNAL OF COMPU TER RESEARCH AND DEV ELOPMEN T Jan 2004 基于分层高斯混合模型的半监督学习算法 孙广玲 　　　唐降龙 ( 哈尔滨工业大学计算机科学与技术学院　哈尔滨 　150001) (laughdoll827 @sinacom) 摘 　要 　提出了一种基于分层高斯混合模型的半监督学习算法 半监督学习算法的学习样本包括已标记类别样本和未 标记类别学习样本 如用高斯混合模型拟合每个类别已标记学习样本的概率分布 ,进而用高斯数为类别数的分层高斯混 ( ) 合模型拟合全部 已标记和未标记 学习样本的分布 ,则形成为一个基于分层的高斯混合模型的半监督学习问题 基于 EM 算法 ,首先利用每个类别已标记样本学习高斯混合模型 ,然后以该模型参数和已标记样本的频率分布作为分层高斯 混合模型参数的初值 ,给出了基于分层高斯混合模型的半监督学习算法 以银行票据印刷体数字识别做实验 ,实验结果 表明 ,本算法能够获得较好的效果 关键词 　半监督学习; 高斯混合模型 ; 分层高斯混合模型 ; EM 算法 中图法分类号 　TP181 A SemiSupervised Learning Algorithm Based on a Hierarchical GMM SUN GuangLing and TAN G XiangLong ( ) School of Computer Science and Technology , Harbin Institute of Technology , Harbin 150001 Abstract 　A semisupervised learning algorithm based on a hierarchical GMM is proposed The learning samples in semisupervised learning are a hybrid of labeled and unlabeled samples If GMM is employed to fit the distribution of labeled samples in each class and a hierarchical GMM whose Gaussian number is the class number is employed to fit the distribution of the whole learning samples (including labeled and unla beled samples) , then a semisupervised learning problem based on a hierarchical GMM has emerged Based on EM algorithm , by learning the labeled samples of each class , a GMM is obtained first Then by taking the parameters of the obtained GMM and frequencies of labeled samples as initials , a semisupervised learn ing algorithm based on a hierarchical GMM is presented Printed numerals in a bank check are tested in the experiments and the results shows the good effects of the propose