Deakin University
Browse

File(s) under permanent embargo

MCNC: multi-channel nonparametric clustering from heterogeneous data

Version 2 2024-06-05, 11:50
Version 1 2017-05-25, 22:37
conference contribution
posted on 2024-06-05, 11:50 authored by TB Nguyen, V Nguyen, Svetha VenkateshSvetha Venkatesh, D Phung
Bayesian nonparametric (BNP) models have recently become popular due to their flexibility in identifying the unknown number of clusters. However, they have difficulties handling heterogeneous data from multiple sources. Existing BNP methods either treat each of these sources independently - hence do not get benefits from the correlating information between them, or require to explicitly specify data sources as primary and context channels. In this paper, we present a BNP framework, termed MCNC, which has the ability to (1) discover co-patterns from multiple sources; (2) explore multi-channel data simultaneously and treat them equally; (3) automatically identify a suitable number of patterns from data; and (4) handle missing data. The key idea is to utilize a richer base measure of a BNP model being a product-space. We demonstrate our framework on synthetic and real-world datasets to discover the identity-location-time (a.k.a who-where-when) patterns. The experimental results highlight the effectiveness of our MCNC framework in both cases of complete and missing data.

History

Pagination

3633-3638

Location

Cancun, Mexico

Start date

2016-12-04

End date

2016-12-08

ISSN

1051-4651

ISBN-13

9781509048472

Language

eng

Publication classification

E Conference publication, E1 Full written paper - refereed

Copyright notice

2016, by the Institute of Electrical and Electronics Engineers, Inc

Editor/Contributor(s)

[Unknown]

Title of proceedings

2016 23rd International Conference on Pattern Recognition (ICPR 2016)

Event

Pattern Recognition. Conference (23rd : 2016 : Cancun, Mexico)

Publisher

IEEE

Place of publication

Piscataway, N.J.