3633 12 2335 4 10973 document 2335 tr04a.pdf application/pdf 146387 2009-12-08 20:48:27 http://cogprints.org/3633/1/tr04a.pdf 3633 1 application/pdf application/pdf en public

tr04a.pdf

http://eprints.org/relation/hasVolatileVersion http://cogprints.org/id/document/5892 http://eprints.org/relation/haspreviewThumbnailVersion http://cogprints.org/id/document/5892 http://eprints.org/relation/hasVersion http://cogprints.org/id/document/5892 archive 4893 disk0/00/00/36/33 2004-05-14 2011-03-11 08:55:36 2007-09-12 16:52:28 techreport show 0 Many existing independent component analysis algorithms include a preprocessing stage where the inputs are sphered. This amounts to normalising the data such that all correlations between the variables are removed. In this work, I show that sphering allows very weak contextual modulation to steer the development of meaningful features. Context-biased competition has been proposed as a model of covert attention and I propose that sphering-like normalisation also allows weaker top-down bias to guide attention. Valpola Harri Artificial Intelligence Laboratory University of Zurich pub FALSE FALSE comp-neuro-sci comp-sci-mach-learn comp-sci-neural-nets comp-sci-art-intel published 2004-05 public