--- abstract: "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.\n" altloc: [] chapter: ~ commentary: ~ commref: ~ confdates: ~ conference: ~ confloc: ~ contact_email: ~ creators_id: [] creators_name: - family: Valpola given: Harri honourific: '' lineage: '' date: 2004-05 date_type: published datestamp: 2004-05-14 department: Artificial Intelligence Laboratory dir: disk0/00/00/36/33 edit_lock_since: ~ edit_lock_until: ~ edit_lock_user: ~ editors_id: [] editors_name: [] eprint_status: archive eprintid: 3633 fileinfo: /style/images/fileicons/application_pdf.png;/3633/1/tr04a.pdf full_text_status: public importid: ~ institution: University of Zurich isbn: ~ ispublished: pub issn: ~ item_issues_comment: [] item_issues_count: 0 item_issues_description: [] item_issues_id: [] item_issues_reported_by: [] item_issues_resolved_by: [] item_issues_status: [] item_issues_timestamp: [] item_issues_type: [] keywords: ~ lastmod: 2011-03-11 08:55:36 latitude: ~ longitude: ~ metadata_visibility: show note: ~ number: ~ pagerange: ~ pubdom: FALSE publication: ~ publisher: ~ refereed: FALSE referencetext: ~ relation_type: [] relation_uri: [] reportno: ~ rev_number: 12 series: ~ source: ~ status_changed: 2007-09-12 16:52:28 subjects: - comp-neuro-sci - comp-sci-mach-learn - comp-sci-neural-nets - comp-sci-art-intel succeeds: ~ suggestions: ~ sword_depositor: ~ sword_slug: ~ thesistype: ~ title: Behaviourally meaningful representations from normalisation and context-guided denoising type: techreport userid: 4893 volume: ~