Sparse Gaussian Graphical Models for Speech Recognition

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract / Description of output

We address the problem of learning the structure of Gaussian graphical models for use in automatic speech recognition, a means of controlling the form of the inverse covariance matrices of such systems. With particular focus on data sparsity issues, we implement a method for imposing graphical model structure on a Gaussian mixture system, using a convex optimisation technique to maximise a penalised likelihood expression. The results of initial experiments on a phone recognition task show a performance improvement over an equivalent full-covariance system.
Original languageEnglish
Title of host publicationInterspeech 2007
Subtitle of host publication8th Annual Conference of the International Speech Communication Association
Pages2113-2116
Number of pages4
Publication statusPublished - 1 Aug 2007

Fingerprint

Dive into the research topics of 'Sparse Gaussian Graphical Models for Speech Recognition'. Together they form a unique fingerprint.

Cite this