0.25
0.5
0.75
1.25
1.5
1.75
2
Experiments with Non-parametric Topic Models
Published on Oct 07, 20142069 Views
In topic modelling, various alternative priors have been developed, for instance asymmetric and symmetric priors for the document-topic and topic-word matrices respectively, the hierarchical Dirichlet
Related categories
Chapter list
Experiments with Non-parametric Topic Models00:00
We are hiring!00:11
Outline - 100:21
Topic Models - 100:25
Topic Models - 200:45
Using Topic Models00:56
High Fidelity Topic Models and Visualisation01:21
Text and Burstiness01:43
Previous Work03:10
Better Sampling Methods for HDP and HPYP03:23
Outline - 204:22
Evolution of Models - 104:27
Evolution of Models - 204:36
Evolution of Models - 305:01
Evolution of Models - 405:13
Evolution of Models - 505:49
Our Non-parametric Topic Model06:17
Handling Burstiness06:22
Outline - 307:13
Runtime Characteristics07:36
Performance Metrics08:55
Performance on Reuters-21578 ModLewis Split09:26
Comparison with Mallet10:16
Comparison with Sato, Kurihawa, Nakagawa, KDD 201211:13
Comparison to Bryant+Sudderth (2012) on NIPS data11:51
Effect of Hyperparameters on the Number of Topics13:12
Outline - 414:34
Conclusion - 114:38
Conclusion - 215:02
Conclusion - 315:24
Conclusion - 415:39
Conclusion - 515:46
Thank You15:58