Home // DBKDA 2017, The Ninth International Conference on Advances in Databases, Knowledge, and Data Applications // View article

A Pseudometric for Gaussian Mixture Models

Authors:
Linfei Zhou
Wei Ye
Bianca Wackersreuther
Claudia Plant
Chrisitian Boehm

Keywords: Gaussian Mixture Models, Similarity Measures, Metric

Abstract:
Efficient similarity search for uncertain data is a challenging task in many modern data mining applications such as image retrieval, speaker recognition and stock market analysis. A common way to model the uncertainty of the objects is using probability density functions in the form of Gaussian Mixture Models (GMMs), which have the ability to approximate any arbitrary distribution. However, there is a lack of suitable similarity measures for GMMs. Hence, in this paper we propose a similarity measure, Pseudometric for GMMs (PmG). The advantage of PmG is that it is efficient in computation because of its closed-form expression for GMMs, and it fulfills the triangle inequality which is necessary for many techniques like clustering and embedding. Extensive experimental evaluations of the proposed similarity measure on various real-world and synthetic data sets demonstrate a considerably better performance than that of the existing similarity measures, in terms of run-time and result quality in classification and clustering.

Pages: 37 to 42

Copyright: Copyright (c) IARIA, 2017

Publication date: May 21, 2017

Published in: conference

ISSN: 2308-4332

ISBN: 978-1-61208-558-6

Location: Barcelona, Spain

Dates: from May 21, 2017 to May 25, 2017