Weipeng Huang (University College Dublin)

will speak on

Variational Bayesian Hierarchical Mixture Clustering and Regularisation

Time: 12:00PM
Date: Mon 30th September 2019
Location: Seminar Room SCN 1.25 [map]

Abstract: Hierarchical clustering is a task that aims for building hierarchies for a collection of items, and its applications are commonly seen in daily life. This fact motivates us to keep seeking better methods, in particular, those which not only maintain theoretical guarantee but are also capable of handling extensive data. This work continues developing the statistical approach, Bayesian Hierarchical Mixture Clustering (BHMC), which constructs the hierarchies with a generative process tackling the data with complex latent structures. To further enhance the structure of the generated hierarchies, we propose a regularisation to the posterior of the BHMC model. Besides, we implement a variational Bayes (VB) approach since VB is scalable and can integrate with the regularised posterior trivially.

(This talk is part of the Working Group on Statistical Learning series.)

PDF notice

Return to all seminars