Difference between revisions of "Eisenstein et al 2011: Sparse Additive Generative Models of Text"
Line 12: | Line 12: | ||
== Brief Description of the method == | == Brief Description of the method == | ||
− | This paper first | + | This paper first describes three big disadvantages of [[UsesMethod::Latent Dirichlet Allocation]]: high inference cost, overparameterization, and lack of sparsity representation. Then, it introduces SAGE, an additive generative model which does not require learning the same background distribution again and again, but rather introduces a sparse topic model that performs addition in log-space. |
− | |||
− | === | + | === The Generative Story === |
+ | |||
+ | === Parameter Estimation === | ||
== Dataset and Experiment Settings == | == Dataset and Experiment Settings == |
Revision as of 19:48, 28 November 2011
Contents
Citation
Sparse Additive Generative Models of Text. Eisenstein, Ahmed and Xing. Proceedings of ICML 2011.
Online version
Summary
This recent paper presents sparse learning and additive generative modeling approaches for Topic modeling. This is an important alternative approach to Latent Dirichlet Allocation (LDA) where sparsity and log-space additive modeling are NOT considered or introduced.
Brief Description of the method
This paper first describes three big disadvantages of Latent Dirichlet Allocation: high inference cost, overparameterization, and lack of sparsity representation. Then, it introduces SAGE, an additive generative model which does not require learning the same background distribution again and again, but rather introduces a sparse topic model that performs addition in log-space.
The Generative Story
Parameter Estimation
Dataset and Experiment Settings
[[File:]]
Experimental Results
The authors performed three major experiments. The first experiment is the . The second experiment explores.
Exp
[[File:]]
Exp
[[File:]]
Exp
[[File:]]
Related Papers
This paper is related to many papers in three dimensions.
(1) .
(2) .
(3) .