Conference item icon

Conference item

Attribute based shared hidden layers for cross-language knowledge transfer

Abstract:

Deep neural network (DNN) acoustic models can be adapted to under-resourced languages by transferring the hidden layers. An analogous transfer problem is popular as few-shot learning to recognise scantily seen objects based on their meaningful attributes. In similar way, this paper proposes a principled way to represent the hidden layers of DNN in terms of attributes shared across languages. The diverse phoneme sets of different languages can be represented in terms of phonological features t...

Expand abstract
Publication status:
Published
Peer review status:
Peer reviewed

Actions


Access Document


Files:
  • (Accepted manuscript, pdf, 270.3KB)
Publisher copy:
10.1109/SLT.2016.7846327

Authors


More by this author
Institution:
University of Oxford
Division:
HUMS
Department:
Linguistics Philology and Phonetics Faculty
Role:
Author
More by this author
Institution:
University of Oxford
Division:
HUMS
Department:
Linguistics Philology and Phonetics Faculty
Role:
Author
More from this funder
Grant:
Proof of Concept FLEX-SR award no. 632226
Publisher:
Institute of Electrical and Electronics Engineers Publisher's website
Journal:
IEEE Workshop on Spoken Language Technology Journal website
Host title:
SLT 2016: IEEE Workshop on Spoken Language Technology
Publication date:
2017-02-09
Acceptance date:
2016-09-16
Event location:
San Diego
Event start date:
2016-12-13T00:00:00Z
Event end date:
2016-12-16T00:00:00Z
DOI:
Keywords:
Pubs id:
pubs:664070
UUID:
uuid:a6db756d-e151-4cd6-804a-9c95b5f425cd
Local pid:
pubs:664070
Deposit date:
2016-12-03

Terms of use


Views and Downloads






If you are the owner of this record, you can report an update to it here: Report update to this record

TO TOP