Deakin University
Browse

File(s) under permanent embargo

Small-variance asymptotics for bayesian nonparametric models with constraints

Version 2 2024-06-06, 01:30
Version 1 2015-08-26, 14:49
chapter
posted on 2024-06-06, 01:30 authored by C Li, Santu RanaSantu Rana, Q Phung, Svetha VenkateshSvetha Venkatesh
The users often have additional knowledge when Bayesian nonparametric models (BNP) are employed, e.g. for clustering there may be prior knowledge that some of the data instances should be in the same cluster (must-link constraint) or in different clusters (cannot-link constraint), and similarly for topic modeling some words should be grouped together or separately because of an underlying semantic. This can be achieved by imposing appropriate sampling probabilities based on such constraints. However, the traditional inference technique of BNP models via Gibbs sampling is time consuming and is not scalable for large data. Variational approximations are faster but many times they do not offer good solutions. Addressing this we present a small-variance asymptotic analysis of the MAP estimates of BNP models with constraints. We derive the objective function for Dirichlet process mixture model with constraints and devise a simple and efficient K-means type algorithm. We further extend the small-variance analysis to hierarchical BNP models with constraints and devise a similar simple objective function. Experiments on synthetic and real data sets demonstrate the efficiency and effectiveness of our algorithms.

History

Volume

9078

Chapter number

8

Pagination

92-105

Location

Vietnam

Start date

2015-01-01

End date

2015-01-01

ISSN

0302-9743

eISSN

1611-3349

ISBN-13

9783319180380

Language

eng

Publication classification

B Book chapter, B1 Book chapter

Copyright notice

2015, Springer

Extent

59

Editor/Contributor(s)

Cao T, Lim EP, Zhou ZH, Ho TB, Cheung D, Motoda H

Publisher

Springer

Place of publication

Berlin, Germany

Title of book

Advances in knowledge discovery and data mining 19th Pacific-Asia Conference, PAKDD 2015, Ho Chi Minh City, Vietnam, May 19-22, 2015, Proceedings, Part II

Series

Lecture notes in computer science; v.9078

Usage metrics

    Research Publications

    Categories

    No categories selected

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC