Edinburgh Research Explorer

Regularization of context-dependent deep neural networks with context-independent multi-task training

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

Original languageEnglish
Title of host publicationProc IEEE International Conference on Acoustics, Speech and Signal Processing
Place of PublicationBrisbane, QLD, Australia
PublisherInstitute of Electrical and Electronics Engineers (IEEE)
Pages4290-4294
Number of pages5
ISBN (Electronic)978-1-4673-6997-8
DOIs
Publication statusPublished - 6 Aug 2015
Event40th IEEE International Conference on Acoustics, Speech and Signal Processing - Brisbane Convention & Exhibition Centre, Brisbane, Australia
Duration: 19 Apr 201524 Apr 2015

Conference

Conference40th IEEE International Conference on Acoustics, Speech and Signal Processing
Abbreviated titleICASSP 2015
CountryAustralia
CityBrisbane
Period19/04/1524/04/15

Abstract

The use of context-dependent targets has become standard in hybrid DNN systems for automatic speech recognition. However, we argue that despite the use of state-tying, optimising to context-dependent targets can lead to over-fitting, and that discriminating between arbitrary tied context-dependent targets may not be optimal. We propose a multitask learning method where the network jointly predicts context-dependent and monophone targets. We evaluate the method on a large-vocabulary lecture recognition task and show that it yields relative improvements of 3-10% over baseline systems.

Event

40th IEEE International Conference on Acoustics, Speech and Signal Processing

19/04/1524/04/15

Brisbane, Australia

Event: Conference

Download statistics

No data available

ID: 19940548