Capsule Networks with Routing Annealing

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Capsule Networks overcome some shortcomings of convolutional neural networks organizing neurons into groups of capsules. Capsule layers are dynamically connected by means of an iterative routing mechanism, which models the connection strengths between capsules from different layers. However, whether routing improves the network performance is still object of debate. This work tackles this issue via Routing Annealing (RA), where the number of routing iterations is annealed at training time. This proposal gives some insights on the effectiveness of the routing for Capsule Networks. Our experiments on different datasets and architectures show that RA yields better performance over a reference setup where the number of routing iterations is fixed (even in the limit case with no routing), especially for architectures with fewer parameters.

Original languageEnglish
Title of host publicationArtificial Neural Networks and Machine Learning – ICANN 2021 - 30th International Conference on Artificial Neural Networks, Proceedings
EditorsIgor Farkaš, Paolo Masulli, Sebastian Otte, Stefan Wermter
PublisherSpringer Science and Business Media Deutschland GmbH
Pages529-540
Number of pages12
ISBN (Print)9783030863616
DOIs
Publication statusPublished - 1 Jan 2021
Externally publishedYes
Event30th International Conference on Artificial Neural Networks, ICANN 2021 - Virtual, Online, Slovakia
Duration: 14 Sept 202117 Sept 2021

Publication series

NameLecture Notes in Computer Science
Volume12891 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference30th International Conference on Artificial Neural Networks, ICANN 2021
Country/TerritorySlovakia
CityVirtual, Online
Period14/09/2117/09/21

Keywords

  • Annealing
  • Capsule networks
  • Routing

Fingerprint

Dive into the research topics of 'Capsule Networks with Routing Annealing'. Together they form a unique fingerprint.

Cite this