Conference item icon

Conference item

Trusting SVM for Piecewise Linear CNNs

Abstract:

We present a novel layerwise optimization algorithm for the learning objective of Piecewise-Linear Convolutional Neural Networks (PL-CNNs), a large class of convolutional neural networks. Specifically, PL-CNNs employ piecewise linear non-linearities such as the commonly used ReLU and max-pool, and an SVM classifier as the final layer. The key observation of our approach is that the problem corresponding to the parameter estimation of a layer can be formulated as a difference-of-convex (DC) pr...

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

Actions


Access Document


Files:
  • (Accepted manuscript, pdf, 650.1KB)

Authors


More by this author
Institution:
University of Oxford
Division:
MPLS
Department:
Engineering Science
Role:
Author
More by this author
Institution:
University of Oxford
Division:
MPLS
Department:
Engineering Science
Role:
Author
More by this author
Institution:
University of Oxford
Oxford college:
Lady Margaret Hall
Role:
Author
More from this funder
Grant:
EP/M013774/1/EPSRC Programme Grant Seebibyte
Journal:
International Conference on Learning Representations Journal website
Host title:
5th International Conference on Learning Representations (ICLR 2017)
Publication date:
2017-04-26
Acceptance date:
2017-02-06
Event location:
Palais des Congrès Neptune, Toulon, France
Event start date:
2017-04-24T00:00:00Z
Event end date:
2017-04-26T00:00:00Z
Source identifiers:
688986
Pubs id:
pubs:688986
UUID:
uuid:08f4a60f-b5f7-40a5-aff9-7ce4bce85b5b
Local pid:
pubs:688986
Deposit date:
2017-04-11

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