Conference item icon

Conference item

Using synthetic data to train neural networks is model-based reasoning

Abstract:

We draw a formal connection between using synthetic training data to optimize neural network parameters and approximate, Bayesian, model-based reasoning. In particular, training a neural network using synthetic data can be viewed as learning a proposal distribution generator for approximate inference in the synthetic-data generative model. We demonstrate this connection in a recognition task where we develop a novel Captcha-breaking architecture and train it using synthetic data, demonstratin...

Expand abstract
Publication status:
Published
Peer review status:
Peer reviewed
Version:
Accepted manuscript

Actions


Access Document


Files:
Publisher copy:
10.1109/IJCNN.2017.7966298

Authors


More by this author
Department:
Oxford, MPLS, Engineering Science
Role:
Author
More by this author
Department:
Oxford, MPLS, Engineering Science
Role:
Author
More by this author
Department:
Kellogg College
Role:
Author
More from this funder
Funding agency for:
Le, TA
Publisher:
IEEE Publisher's website
Pages:
3514-3521
Publication date:
2017-07-03
Acceptance date:
2017-02-27
DOI:
EISBN:
978-1-5090-6182-2
ISSN:
2161-4407
Pubs id:
pubs:684943
URN:
uri:d1bcfc62-a8f0-4e9a-8301-927a3be56af9
UUID:
uuid:d1bcfc62-a8f0-4e9a-8301-927a3be56af9
Local pid:
pubs:684943
ISBN:
978-1-5090-6183-9

Terms of use


Metrics


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