Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • News & Views
  • Published:

Computational neuroscience

Think positive to find parts

Nature volume 401, pages 759–760 (1999)Cite this article

How do you tell a shovel from an acorn? A bicycle from a mango? A young face from an old one? They look different, of course, but what exactly are the 'features'or 'parts' that our brains use to tell one thing from another? On page 788 of this issue, Lee and Seung1 present a new algorithm for learning object parts. Known as non-negative matrix factorization (NMF), their algorithm follows from the simple prescription that parts are collections of input elements that tend to come and go together, and which can be added together in different combinations to give objects of interest.

In the visual realm, the authors' view of parts is similar to the idea behind identikits — the collections of mix-and-match face parts used to reconstruct the identities of wanted fugitives or lost children. Parts that are too much like whole objects are rejected, because this would require an impracticably large inventory. Lee and Seung's approach also shuns parts which, although few in number, must be combined in counterintuitive mixtures to reconstruct objects of interest.

This is a preview of subscription content, access via your institution

Relevant articles

Open Access articles citing this article.

Access options

Buy this article

  • Purchase on Springer Link
  • Instant access to full article PDF
Buy now

Prices may be subject to local taxes which are calculated during checkout

References

  1. Lee, D. D. & Seung, H.S. Nature 401, 788–791 (1999).

    Article  ADS  CAS  Google Scholar 

  2. Bishop, C. M. Neural Networks for Pattern Recognition (Oxford Univ. Press, 1995).

  3. Hinton, G. E. & Sejnowski, T. J. Unsupervised Learning: Foundations of Neural Computation (MIT Press, Cambridge, Massachusetts, 1999).

  4. Penev, P. S. & Atick, J. J. Comp. Neural Syst. 7, 477–500 (1996).

    Article  Google Scholar 

  5. Biederman, I. Psychol. Rev. 94, 115–147 (1987).

    Article  Google Scholar 

Download references

Author information

Authors and Affiliations

  1. the Department of Biomedical Engineering, University of Southern California, MC 1451, Los Angeles, 90089, California, USA

    Bartlett W. Mel

Authors
  1. Bartlett W. Mel
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Bartlett W. Mel.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Mel, B. Think positive to find parts. Nature 401, 759–760 (1999). https://doi.org/10.1038/44507

Download citation

  • Issue Date:

  • DOI: https://doi.org/10.1038/44507

This article is cited by

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing