Conference item icon

Conference item

Deterministic binary filters for convolutional neural networks

Abstract:

We propose Deterministic Binary Filters, an approach to Convolutional Neural Networks that learns weighting coefficients of predefined orthogonal binary basis instead of the conventional approach of learning directly the convolutional filters. This approach results in model architectures with significantly fewer parameters (4x to 16x) and smaller model sizes (32x due to the use of binary rather than floating point precision). We show our deterministic filter design can be integrated into well...

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

Actions


Access Document


Files:
  • (Version of record, pdf, 509.8KB)
Publisher copy:
10.24963/ijcai.2018/380

Authors


More by this author
Institution:
University of Oxford
Division:
MPLS Division
Department:
Computer Science
Role:
Author
More by this author
Institution:
University of Oxford
Oxford college:
Linacre College
Role:
Author
More by this author
Institution:
University of Oxford
Role:
Author
Expand authors...
Publisher:
International Joint Conferences on Artificial Intelligence Organization Publisher's website
Journal:
IJCAI International Joint Conference on Artificial Intelligence
Pages:
2739-2747
Host title:
Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence
Publication date:
2018-07-19
Acceptance date:
2018-04-16
DOI:
ISSN:
1045-0823
Source identifiers:
930482
ISBN:
9780999241127
Keywords:
Pubs id:
pubs:930482
UUID:
uuid:434cdb31-4fb2-4b6b-8cd6-9139ff067a5c
Local pid:
pubs:930482
Deposit date:
2019-02-26

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