feature detectors

[Juergen Schmidhuber]

           SEMILINEAR PREDICTABILITY MINIMIZATION 
           PRODUCES WELL-KNOWN  FEATURE DETECTORS
      (9 pages, 260 K compressed, 1.14 M uncompressed)
             Neural Computation, 1996 (accepted) 

   Juergen Schmidhuber, Martin Eldracher, Bernhard Foltin 

Predictability minimization (PM) exhibits various intuitive and 
theoretical advantages over many other methods for unsupervised 
redundancy reduction. So far, however, there were only toy appli-
cations of PM. In this paper, we apply semilinear PM to static 
real world images and find: without a teacher and without any 
significant pre-processing, the system automatically learns to 
generate distributed representations based on well-known feature 
detectors, such as orientation sensitive edge detectors and off-
center-on-surround-like structures, thus extracting simple features
related to those considered useful for image pre-processing and
compression. (Revised and extended TR FKI-201-94)

To obtain a copy, cut and paste this:
netscape ftp://ftp.idsia.ch/pub/juergen/detectors.ps.gz

Juergen Schmidhuber, IDSIA
Martin Eldracher, IDSIA / TUM
Bernhard Foltin, TUM