| LDA模型参数有偏估计方法 |
| |
| 引用本文： | 袁伯秋,周一民,李林.LDA模型参数有偏估计方法[J].小型微型计算机系统,2010,31(4). |
| |
| 作者姓名： | 袁伯秋 周一民 李林 |
| |
| 作者单位： | 北京航空航天大学,计算机学院,北京,100191 |
| |
| 基金项目： | 国家"九七三"重点基础研究发展计划项目 |
| |
| 摘 要： | LDA(Latent Dirichlet Allocation)等基于隐含topic的模型在离散数据处理中的应用逐渐增多.然而LDA使用Dirichlet分布作为隐含topic的分布函数,未能很好表示各topic之间相互关系.目前常见改进方法是通过DAG(Directed Acyclic Graph)图或对数正态分布等其他分布函数表达topic之间的关系.本文通过参数有偏估计的方法,考虑topic混合过程中词项上的重叠关系,改变topic内部词项分布,最终改进LDA模型性能.在回顾一些基础内容后,重点介绍参数有偏估计及简化计算方法.最后通过LDA模型在信息检索中的实验验证这种改进的有效性,并初步分析模型参数选用规律.
|
| 关 键 词： | topic模型 参数有偏估计 |
Biased Parameter Estimation for LDA |
| |
| YUAN Bo-qiu,ZHOU Yi-ming,LI Lin.Biased Parameter Estimation for LDA[J].Mini-micro Systems,2010,31(4). |
| |
| Authors: | YUAN Bo-qiu ZHOU Yi-ming LI Lin |
| |
| Affiliation: | YUAN Bo-qiu,ZHOU Yi-ming,LI Lin(School of Computer Science , Technology,Beijing University of Aeronautics , Astronautics,Beijing 100191,China) |
| |
| Abstract: | Latent Dirichlet Allocation(LDA)and other related topic models are increasingly popular tools for applications in discrete data.However,the Dirichlet distribution over latent topics in LDA did not capture the co-relation between topics very well.To overcome the drawback,directed acyclic graph(DAG)and other algebra distribution,such as logistic normal distribution,were introduced to describe the correlations between topics.They were effective but relatively expensive.We alleviated it by bias parameter estima... |
| |
| Keywords: | LDA Wordnet |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
