无限狄利克雷混合模型的变分学习  

Variational Learning for Infinite Dirichlet Mixture Model

作　　者：曾凡锋[1] 陈可[1] 王宝成[1] 肖珂[1] 

机构地区：[1]北方工业大学

出　　处：《计算机技术与发展》2017年第10期19-23,共5页Computer Technology and Development

基　　金：国家自然科学基金资助项目(61371142);北方工业大学校内专项(XN060)

摘　　要：有限高斯混合模型广泛应用于模式识别、机器学习和数据挖掘等领域,但现实中的许多数据都具有非高斯性,而高斯混合模型无法准确地描述这些数据。此外,有限高斯混合模型还存在参数估计和模型选择困难的问题。为了更好地拟合非高斯数据,解决有限高斯混合模型的参数估计和模型选择困难的问题,在研究一种适合于建模非高斯数据的无限狄利克雷混合模型的学习方法的基础上,提出了一种高效的变分近似推理算法。该算法能够同时解决参数估计及模型选择的问题。为了验证该算法的有效性,在合成数据集上进行了大量实验。验证实验结果表明,该算法能够很好地解决模型选择及参数估计的问题。所提出的无限狄利克雷混合模型还可应用于目标检测、文本分类、图像分类等方面。Finite Gauss mixture model is widely used in pattern recognition, machine learning and data mining and so on, but many data in reality are non Gauss, which cannot accurately describe these data. In addition, there exist difficulties in parameter estimation and model selection in the finite Gauss mixture model. In order to better fit the non Gauss data and solve the problem of parameter estimation and model selection of the finite Gauss mixture model, on the basis of research on basic learning method of infinite Dirichlet mixture model suitable for modeling the data of a non Gauss, an efficient variational approximate inference algorithm is proposed, which solves problem of parameter estimation and model selection at the same time. In order to verify its validity, a lot of experiments are carded out on the syn- thetic data set. The experimental results show it can solve the problem of model selection and parameter estimation. Infinite Dirichlet mix- ture model proposed can also be applied to object detection, text classification,image classification and other parts.

关 键 词：狄利克雷 无限混合模型 变分贝叶斯 模型选择 参数估计 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]