Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
The problems with using STNs to align CNN feature maps
KTH, School of Electrical Engineering and Computer Science (EECS), Computer Science, Computational Science and Technology (CST). (Computational Brain Science Lab)ORCID iD: 0000-0001-8548-5788
KTH, School of Electrical Engineering and Computer Science (EECS), Computer Science, Computational Science and Technology (CST). (Computational Brain Science Lab)ORCID iD: 0000-0003-0011-6444
KTH, School of Electrical Engineering and Computer Science (EECS), Computer Science, Computational Science and Technology (CST). (Computational Brain Science Lab)ORCID iD: 0000-0002-9081-2170
2020 (English)Conference paper, Oral presentation with published abstract (Other academic)
Abstract [en]

Spatial transformer networks (STNs) were designed to enable CNNs to learn invariance to image transformations. STNs were originally proposed to transform CNN feature maps as well as input images. This enables the use of more complex features when predicting transformation parameters. However, since STNs perform a purely spatial transformation, they do not, in the general case, have the ability to align the feature maps of a transformed image and its original. We present a theoretical argument for this and investigate the practical implications, showing that this inability is coupled with decreased classification accuracy. We advocate taking advantage of more complex features in deeper layers by instead sharing parameters between the classification and the localisation network.

Place, publisher, year, edition, pages
2020.
Keywords [en]
deep learning, convolutional neural networks, spatial transformer networks, invariant neural networks
National Category
Computer Vision and Robotics (Autonomous Systems)
Research subject
Computer Science
Identifiers
URN: urn:nbn:se:kth:diva-266471OAI: oai:DiVA.org:kth-266471DiVA, id: diva2:1388075
Conference
Northern Lights Deep Learning Workshop 2020, Tromsø, Norway, 20-21 Jan 2020
Funder
Swedish Research Council, 2018-03586
Note

QC 20200123

Available from: 2020-01-23 Created: 2020-01-23 Last updated: 2020-01-24Bibliographically approved

Open Access in DiVA

STN_problems_to_align_CNN_feature_maps(774 kB)7 downloads
File information
File name FULLTEXT02.pdfFile size 774 kBChecksum SHA-512
f6807aa5686d80596fca60eeeab992a1c26d649fd4bfff13d8e263aac45388a6db538db5579ec69c6c729622935193693d7c894f178b03d0e07663d8fabd06fb
Type fulltextMimetype application/pdf

Other links

Conference websiteConference programPreprint in arXiv

Search in DiVA

By author/editor
Finnveden, LukasJansson, YlvaLindeberg, Tony
By organisation
Computational Science and Technology (CST)
Computer Vision and Robotics (Autonomous Systems)

Search outside of DiVA

GoogleGoogle Scholar
Total: 7 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 230 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf