基于协方差的高斯混合模型参数学习算法

计算机科学 ›› 2013, Vol. 40 ›› Issue (Z11): 77-81.

基于协方差的高斯混合模型参数学习算法

廖晓锋,范修斌,姜青山

南昌大学信息工程学院南昌330031;中国科学院软件研究所北京100190;中国科学院深圳先进技术研究院深圳550085

出版日期:2018-11-16 发布日期:2018-11-16
基金资助:
本文受深圳市战略性新兴产业发展专项资金基础研究重点项目:海量恶意软件鉴别关键技术及其在钓鱼网站检测中的应用(JCYJ20120617120716224),江西省教育厅青年科学基金项目:双模态概率主题模型及基于DOT的并行扩展研究(GJJ13013)资助

Covariance Based Learning Algorithm for Gaussian Mixture Model

LIAO Xiao-feng,FAN Xiu-bin and JIANG Qing-shan

Online:2018-11-16 Published:2018-11-16

摘要/Abstract

摘要： 对混合高斯模型参数估计问题的算法通常是基于期望最大(Expectation Maximization)给出的。在混合高斯模型的因素协方差矩阵已知、因素各分量独立的前提下,给出了基于协方差矩阵的机器学习算法,简称CVB(Covariance Based)算法,并进行了一定的数学分析。最后给出了与期望最大算法的实验结果比较。实验结果表明,在该条件下,基于协方差的算法优于期望最大算法。

关键词: 混合高斯模型,期望最大化,协方差,CVB算法

Abstract: Expectation maximization is commonly used for parameter estimation in Gaussian mixture model．This paper presented a machine learning algorithm based on covariance(CVB) for solving the Gaussian mixture model with the specific constrain that covariance is already known．Experiments show that the CVB algorithm has better performance than the EM algorithm with regard to the specific constraint.

Key words: Gaussian mixture model, Expectation maximization,Covariance based,CVB algorithm

廖晓锋,范修斌,姜青山. 基于协方差的高斯混合模型参数学习算法[J]. 计算机科学, 2013, 40(Z11): 77-81. https://doi.org/

LIAO Xiao-feng,FAN Xiu-bin and JIANG Qing-shan. Covariance Based Learning Algorithm for Gaussian Mixture Model[J]. Computer Science, 2013, 40(Z11): 77-81. https://doi.org/

参考文献

[1] Martinez B,Binefa X,Pantic M．Facial component detection in thermal imagery[C]∥Proceedings of the Computer Vision and Pattern Recognition Workshops (CVPRW),2010IEEE Computer Society Conference．June 2010:48-54
[2] McKenna S J,Gong Shao-gang,Raja Y．Modelling Facial Colour and Identity with Gaussian Mixtures[J]．Pattern Recognition,1998,31(12):1883-1892
[3] Figueiredo M．Bayesian Image Segmentation Using GaussianField Priors[M]∥ Rangarajan A,Vemuri B,Yuille A．Energy Minimization Methods in Computer Vision and Pattern Recognition．Berlin,Heidelberg:Springer,2005:74-89
[4] Chad C,Serge B,Hayit G,et al．Blobworld:Image Segmentation Using Expectation-Maximization and Its Application to Image Querying[J]．IEEE Transactions on Pattern Analysis and Machine Intelligence,2002,24:1026-1038
[5] Greenspan H,Ruf A,Goldberger J．Constrained Gaussian mixture model framework for automatic segmentation of MR brain images[J]．IEEE Transactions on Medical Imaging,2006,25(9):1233-1245
[6] 向日华,王润生．一种基于高斯混合模型的距离图像分割算法[J]．软件学报,2003,14(7):1250-1257
[7] 陈允杰,张建伟,韦志辉,等．基于高斯混合模型的活动轮廓模型脑MRI分割[J]．计算机研究与发展,2007,9:1595-1603
[8] Reynolds D A,Rose R C．Robust text-independent speaker identification using Gaussian mixture speaker models[J]．IEEE Transactions on Speech and Audio Processing,1995,3(1):72-83
[9] Reynolds D A,Quatieri T F,Dunn R B．Speaker Verification Using Adapted Gaussian Mixture Models[J]．Digital Signal Processing,2000,10(1-3):19-41
[10] 张怡颖,朱小燕,张钹．与文本无关的说话人自适应确认方法[J]．软件学报,2000,11(6):799-803
[11] Zivkovic Z,van der Heijden F．Recursive unsupervised learning of finite mixture models[J]． IEEE Transactions on Pattern Analysis and Machine Intelligence,2004,26(5):651-656
[12] Figueiredo M,Leito J,Jain A．On Fitting Mixture Models[M]∥Hancock E,Pelillo M．Energy Minimization Methods in ComputerVision and Pattern Recognition． Berlin/Heidelberg:Springer,1999:732-732
[13] 王平波,蔡志明,刘旺锁．混合高斯概率密度模型参数的期望最大化估计[J]．声学技术,2007,26(3):5
[14] Dempster A P,Laird N M,Rubin D B．Maximun Likelihoodfrom Incomplete Data via the EM Algorithm[J]．Journal of the Royal Statistical Society,1977,39(1):1-38
[15] Xu Lei,Jordan M I．On Convergence Properties of the EM Algo-rithm for Gaussian Mixtures[J]．Neural Computation,1996,8:129-151
[16] Mitchell T M．Machine Learning[M]．McGraw-Hill,1997

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed