Timezone: »

Dirichlet belief networks for topic structure learning
He Zhao · Lan Du · Wray Buntine · Mingyuan Zhou

Wed Dec 05 02:00 PM -- 04:00 PM (PST) @ Room 210 #4

Recently, considerable research effort has been devoted to developing deep architectures for topic models to learn topic structures. Although several deep models have been proposed to learn better topic proportions of documents, how to leverage the benefits of deep structures for learning word distributions of topics has not yet been rigorously studied. Here we propose a new multi-layer generative process on word distributions of topics, where each layer consists of a set of topics and each topic is drawn from a mixture of the topics of the layer above. As the topics in all layers can be directly interpreted by words, the proposed model is able to discover interpretable topic hierarchies. As a self-contained module, our model can be flexibly adapted to different kinds of topic models to improve their modelling accuracy and interpretability. Extensive experiments on text corpora demonstrate the advantages of the proposed model.

Author Information

He Zhao (Monash University, Australia)
Lan Du (Monash University)
Wray Buntine (Monash University)

Wray Buntine is a full professor at Monash University where he is directory of the Machine Learning Group. He was previously at NICTA in Canberra, Helsinki Institute for Information Technology where he ran a semantic search project, NASA Ames Research Center, University of California, Berkeley, and Google. In the '90s he was involved in a string of startups for both Wall Street and Silicon Valley. He is known for Bayesian machine learning, non-parametrics and document analysis, having been a driving force in the use of ensembling, graphical models, and nonparametric algorithms.

Mingyuan Zhou (University of Texas at Austin)

More from the Same Authors