title: Behaviourally meaningful representations from normalisation and context-guided denoising
creator: Valpola, Harri
subject: Computational Neuroscience
subject: Machine Learning
subject: Neural Nets
subject: Artificial Intelligence
description: 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. 
date: 2004-05
type: Departmental Technical Report
type: NonPeerReviewed
format: application/pdf
identifier: http://cogprints.org/3633/1/tr04a.pdf
identifier:   Valpola, Harri  (2004) Behaviourally meaningful representations from normalisation and context-guided denoising.  [Departmental Technical Report]     
relation: http://cogprints.org/3633/