Visualization of feature evolution during convolutional neural network training

Arjun Punjabi, Aggelos K. Katsaggelos

Research output: Chapter in Book/Report/Conference proceedingConference contribution

8 Scopus citations

Abstract

Convolutional neural networks (CNNs) are a staple in the fields of computer vision and image processing. These networks perform visual tasks with state-of-the-art accuracy; yet, the understanding behind the success of these algorithms is still lacking. In particular, the process by which CNNs learn effective task-specific features is still unclear. This work elucidates such phenomena by applying recent deep visualization techniques during different stages of the training process. Additionally, this investigation provides visual justification to the benefits of transfer learning. The results are in line with previously discussed notions of feature specificity, and show a new facet of a particularly vexing machine learning pitfall: overfitting.

Original languageEnglish (US)
Title of host publication25th European Signal Processing Conference, EUSIPCO 2017
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages311-315
Number of pages5
ISBN (Electronic)9780992862671
DOIs
StatePublished - Oct 23 2017
Event25th European Signal Processing Conference, EUSIPCO 2017 - Kos, Greece
Duration: Aug 28 2017Sep 2 2017

Publication series

Name25th European Signal Processing Conference, EUSIPCO 2017
Volume2017-January

Other

Other25th European Signal Processing Conference, EUSIPCO 2017
Country/TerritoryGreece
CityKos
Period8/28/179/2/17

Keywords

  • Convolutional neural network
  • Deep learning
  • Feature visualization
  • Transfer learning

ASJC Scopus subject areas

  • Signal Processing

Fingerprint

Dive into the research topics of 'Visualization of feature evolution during convolutional neural network training'. Together they form a unique fingerprint.

Cite this