[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]
On Mon, Feb 9, 2009 at 8:34 PM, Anthony Williams wrote: > A few years ago I worked for a patent information company, and I had > to write a program that used Bayesian filtering to categorize patents > (e.g. chemical, mechanical, industrial, etc., but more more nuanced). I'd expect Bayesian filtering to work almost faultlessly in such cases: in an anti-spam context, where a huge effort is made (by spammers) to confuse such filters, they already work surprisingly well. (The 'surprisingly' is a quote from someone who knows a lot more about that kind of stuff than I do.) I can imagine that if you'd apply it very cleverly on images, you might even be able to classify these. Martijn. -- The Mailing List for the Devon & Cornwall LUG http://mailman.dclug.org.uk/listinfo/list FAQ: http://www.dcglug.org.uk/linux_adm/list-faq.html