Reliability-based estimation of the number of noisy features: Application to model-order selection in the union models

Peter Jančovič, Münevver Köküer, Fionn Murtagh

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

4 Citations (Scopus)

Abstract

This paper is concerned with the multi-stream approach in speech recognition. In a given set of feature streams, there may be some features corrupted by noise. Ideally, these features should be excluded from recognition. To achieve this, a-priori knowledge about the identity, including both the number and location, of the noisy features is required. In this paper, we present a method for estimating the number of noisy feature streams. This method assumes no knowledge about the noise. It is based on calculation of the reliability of each feature stream and then evaluation of the joint maximal reliability. Since this method decreases the uncertainty about the noisy features and is statistical in nature, it can also be used to increase robustness of other classification systems. We present an application of this method to model-order selection in the union models. We performed tests on the TIDIGITS database, corrupted by noises affecting various numbers of feature streams. The experimental results show that this model achieves recognition performance similar to the one obtained with a-priori knowledge about the identity of the corrupted features.

Original languageEnglish
Title of host publication2003 IEEE International Conference on Acoustics, Speech, and Signal Processing
Subtitle of host publicationProceedings (ICASSP '03)
EditorsBillene Mercer
PublisherIEEE
Pages416-419
Number of pages4
Volume1
ISBN (Print)9780780376632, 0780376633
DOIs
Publication statusPublished - 21 May 2003
Externally publishedYes
Event2003 IEEE International Conference on Accoustics, Speech, and Signal Processing - Hong Kong, Hong Kong
Duration: 6 Apr 200310 Apr 2003

Publication series

NameIEEE International Conference on Acoustics, Speech and Signal Processing
PublisherIEEE
Volume1
ISSN (Print)1520-6149

Conference

Conference2003 IEEE International Conference on Accoustics, Speech, and Signal Processing
Country/TerritoryHong Kong
CityHong Kong
Period6/04/0310/04/03

Fingerprint

Dive into the research topics of 'Reliability-based estimation of the number of noisy features: Application to model-order selection in the union models'. Together they form a unique fingerprint.

Cite this