Ändra sökning
RefereraExporteraLänk till posten
Permanent länk

Direktlänk
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Annat format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annat språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf
Target aware network adaptation for efficient representation learning
KTH, Skolan för elektroteknik och datavetenskap (EECS), Robotik, perception och lärande, RPL.
KTH, Skolan för elektroteknik och datavetenskap (EECS), Robotik, perception och lärande, RPL.
(Toshiba Corporated R&D Center)
KTH, Skolan för elektroteknik och datavetenskap (EECS), Robotik, perception och lärande, RPL.ORCID-id: 0000-0002-4266-6746
2018 (Engelska)Ingår i: ECCV 2018: Computer Vision – ECCV 2018 Workshops, Munich: Springer, 2018, Vol. 11132, s. 450-467Konferensbidrag, Publicerat paper (Refereegranskat)
Abstract [en]

This paper presents an automatic network adaptation method that finds a ConvNet structure well-suited to a given target task, e.g. image classification, for efficiency as well as accuracy in transfer learning. We call the concept target-aware transfer learning. Given only small-scale labeled data, and starting from an ImageNet pre-trained network, we exploit a scheme of removing its potential redundancy for the target task through iterative operations of filter-wise pruning and network optimization. The basic motivation is that compact networks are on one hand more efficient and should also be more tolerant, being less complex, against the risk of overfitting which would hinder the generalization of learned representations in the context of transfer learning. Further, unlike existing methods involving network simplification, we also let the scheme identify redundant portions across the entire network, which automatically results in a network structure adapted to the task at hand. We achieve this with a few novel ideas: (i) cumulative sum of activation statistics for each layer, and (ii) a priority evaluation of pruning across multiple layers. Experimental results by the method on five datasets (Flower102, CUB200-2011, Dog120, MIT67, and Stanford40) show favorable accuracies over the related state-of-the-art techniques while enhancing the computational and storage efficiency of the transferred model.

Ort, förlag, år, upplaga, sidor
Munich: Springer, 2018. Vol. 11132, s. 450-467
Serie
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), ISSN 0302-9743 ; 11132
Nationell ämneskategori
Datorsystem
Identifikatorer
URN: urn:nbn:se:kth:diva-250561DOI: 10.1007/978-3-030-11018-5_38Scopus ID: 2-s2.0-85061697164ISBN: 9783030110178 (tryckt)OAI: oai:DiVA.org:kth-250561DiVA, id: diva2:1308064
Konferens
15th European Conference on Computer Vision, ECCV 2018; Munich; Germany; 8 September 2018 through 14 September 2018
Anmärkning

QC 20190627

Tillgänglig från: 2019-04-30 Skapad: 2019-04-30 Senast uppdaterad: 2019-06-27Bibliografiskt granskad

Open Access i DiVA

Fulltext saknas i DiVA

Övriga länkar

Förlagets fulltextScopus

Personposter BETA

Li, Vladimir

Sök vidare i DiVA

Av författaren/redaktören
Yang, ZhongLi, VladimirMaki, Atsuto
Av organisationen
Robotik, perception och lärande, RPL
Datorsystem

Sök vidare utanför DiVA

GoogleGoogle Scholar

doi
isbn
urn-nbn

Altmetricpoäng

doi
isbn
urn-nbn
Totalt: 12 träffar
RefereraExporteraLänk till posten
Permanent länk

Direktlänk
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Annat format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annat språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf