Friday, July 10, 2009

Buehler, Zisserman, Everingham Computer Learns Sign Language

"Once the team were confident the computer could identify different signs in this way, they exposed it to around 10 hours of TV footage that was both signed and subtitled. They tasked the software with learning the signs for a mixture of 210 nouns and adjectives that appeared multiple times during the footage."

http://www.newscientist.com/article/dn17431-computer-learns-sign-language-by-watching-tv.html

Patrick Buehler:
http://patrick.buehler.googlepages.com/home

Andrew Zisserman:
http://www.robots.ox.ac.uk/~az/

Mark Everingham:
http://www.comp.leeds.ac.uk/me/

"We propose a framework based on multiple instance
learning which can learn a large number of British Sign
Language signs from TV broadcasts. We achieve very
promising results even under these weak and noisy conditions
by using a state-of-the-art upper-body tracker, descriptors
of the hands that properly model the case of touching
hands, and a plentiful supply of data. A similar method
could be applied to a variety of fields where weak supervision
is available, such as learning gestures and actions."

Learning Sign Language by Watching Tv

Social Security Numbers Can Be Predicted With Public Information

"Carnegie Mellon University researchers have shown that public information readily gleaned from governmental sources, commercial data bases, or online social networks can be used to routinely predict most — and sometimes all — of an individual's nine-digit Social Security number."

http://www.sciencedaily.com/releases/2009/07/090706171509.htm

More info from CMU: http://blogs.heinz.cmu.edu/ssnstudy/

Ralph Gross:
http://www.ri.cmu.edu/person.html?person_id=742

Alessandro Acquisti:
http://www.heinz.cmu.edu/~acquisti/
"Not by strength, by guile."