Multilayer self-organizing category formation networks

Masao Sakai, Noriyasu Homma, Yosuke Koyanaka, Kenichi Abe

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

Abstract

In this paper, we develop a new multilayer neural model that forms categories of inputs for some practical applications such as pattern recognition, learning, image processing, and trend analysis. An essential core of the model is to use a novel vector representation of concepts that compose an input in a multi-level informational hierarchy that makes the model possess category formation ability from incomplete observation of the input. Simulation results demonstrate the usefulness of the model for a facial image recognition task, even if it is carried out under an incremental and unsupervised learning environment. In addition, we evaluate the adequacy and efficiency of formed categories by using principal component analysis.

Original languageEnglish
Title of host publicationProceedings of the SICE Annual Conference
Pages2895-2900
Number of pages6
Publication statusPublished - 2005
EventSICE Annual Conference 2005 - Okayama, Japan
Duration: 2005 Aug 82005 Aug 10

Other

OtherSICE Annual Conference 2005
CountryJapan
CityOkayama
Period05/8/805/8/10

Keywords

  • Concept formation
  • Hebbian rule
  • Incremental learning
  • Neural networks
  • Pattern recognition
  • Principal component analysis
  • Self-organization

ASJC Scopus subject areas

  • Engineering(all)

Fingerprint Dive into the research topics of 'Multilayer self-organizing category formation networks'. Together they form a unique fingerprint.

  • Cite this

    Sakai, M., Homma, N., Koyanaka, Y., & Abe, K. (2005). Multilayer self-organizing category formation networks. In Proceedings of the SICE Annual Conference (pp. 2895-2900)