Variational Learning for Finite Beta-Liouville Mixture Models and Its Application

LAI Yu-ping; DING Hong-wei; ZHOU Ya-jian; GUO Yu-cui; YANG Yi-xian

doi:10.3969/j.issn.0372-2112.2014.07.015

您当前的位置：

首页 >

文章列表页 >

Variational Learning for Finite Beta-Liouville Mixture Models and Its Application

更新时间：2025-07-16

- Variational Learning for Finite Beta-Liouville Mixture Models and Its Application
- Acta Electronica Sinica Vol. 42, Issue 7, Pages: 1347-1352(2014)
- 作者机构：
  
  1. 北京邮电大学计算机学院,北京,100876
  2. 云南大学信息学院,云南,昆明,650091
  3. 北京邮电大学理学院,北京,100876
  4. 北京邮电大学计算机学院,北京,100876
  5. 云南大学信息学院,云南,昆明,650091
  6. 北京邮电大学理学院,北京,100876
- 作者简介：
- 基金信息：
- DOI：10.3969/j.issn.0372-2112.2014.07.015
  CLC： TP391.4
- Published：2014
- 稿件说明：
移动端阅览
LAI Yu-ping, DING Hong-wei, ZHOU Ya-jian, et al. Variational Learning for Finite Beta-Liouville Mixture Models and Its Application[J]. Acta Electronica Sinica, 2014, 42(7): 1347-1352.
DOI：

LAI Yu-ping, DING Hong-wei, ZHOU Ya-jian, et al. Variational Learning for Finite Beta-Liouville Mixture Models and Its Application[J]. Acta Electronica Sinica, 2014, 42(7): 1347-1352. DOI： 10.3969/j.issn.0372-2112.2014.07.015.

摘要

由于贝塔刘维尔分布的共轭先验分布中存在积分表达式，贝叶斯估计有限贝塔刘维尔混合模型参数异常困难.本文提出利用变分贝叶斯学习模型参数，采用gamma分布作为近似的先验分布并使用合理的非线性近似技术，得到了后验分布的近似解.与常用的EM算法相比，该方法能够同时估计模型参数和确定分量数，且避免了过拟合的问题.在合成数据集及场景分类问题上进行了大量的实验，实验结果验证了本文所提方法的有效性.

Abstract

Since the integration expression is present in the conjugate prior distribution

Bayesian estimation of the parameters in finite Beta-Liouville mixture models(BLM)is analytically intractable.In this paper

an approach based on the variational inference framework is proposed.Adopting gamma distributions to approximate the prior distributions of the parameter in BLM and using some reasonable non-linear approximations;the closed form solution for the posterior distribution of the parameters is obtained.Compared to the conventional expectation maximization(EM)algorithm

the proposed algorithm is able to simultaneously estimate the model parameters and determine the number of components;our method also avoids the problem of overfitting.Extensive experimental results based on the synthetic data sets and scenes classification show that the proposed method is efficient and feasible in terms of parameter estimation and model selection.

关键词

Keywords

references

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

High-Range Resolution Radar Target Recognition Based on Deep Interpretable Probabilistic Auto-Encoder Model

Diverse Image Captioning via Conditional Variational Inference and Introspective Adversarial Learning

Traffic Flow Prediction Model Based on Spatio-Temporal Feature Distillation Variational Autoencoder

Unsupervised Learning for Finite Mixture Models Based on BIC Criterion and Gibbs Sampling

A New Model Selection Criterion Based on Kullback-Leibler Information Divergence and Geometric Curvature

Related Author

LIAO Lei-yao

LIU Hao

LIU Ming-ming

LI Sui

LIU Bing

LI Yan-ling

TANG Wen-yan

OUYANG Yi

Related Institution

National Key Laboratory of Radar Signal Processing, Xidian University

College of Communication and Information Engineering, Nanjing University of Posts and Telecommunications

Mine Digitization Engineering Research Center of the Ministry of Education

School of Computer Science and Technology， China University of Mining and Technology

School of Management Science and E-Commerce， Zhejiang Gongshang University

⁰