Re: Facial expression recognition software

Aaron Davidson (ajd@gpu.srv.ualberta.ca)
Thu, 8 Apr 1999 17:57:06 -0600

>> The system's creators said it is as accurate as a trained psychologist and
>> capture and identify subliminal expressions that pass over a face
>>momentarily
>> before a "posed" expression is consciously assumed. That makes the
>>technology
>> potentially applicable as a lie detector.
>
>Despite their success in chess, nobody has yet made a computer that
>can come anywhere close to even a moderately good human in poker.
>Sounds like this bulwark of human skill is falling too. (For those
>of you who believe that poker is a game of chance, please bring
>your money to Sacramento and I will demonstrate otherwise:-)

It is preciscely this challenge which I will be working on this summer. A group of AI researchers at the University of Alberta (including one ex Poker Pro) are attempting to develop a world-class poker playing program.

I'm just an undergrad summer student, so my input will no doubt be limited. It certainly is a tough problem -- the incomplete information, bluffing, and opponent modelling are much trickier tasks than closed information games such as chess.

If any group can do it it will be these guys though -- Dr. Jonathan Schaeffer, the lead researcher, has already conquered checkers with his famous 'chinook' program, the world champion checkers player.

http://www.cs.ualberta.ca/~games/

A module that examined camera-feed images to analyze opponent's poker-faces would certainly be an advantage ;-). The computer has the ultimate poker-face.

--Aaron 

+-----------------------------------------------------------------------+
|  Aaron Davidson          |  <ajd@gpu.srv.ualberta.ca>                 |
| Silicon Creek Software | <http://ugweb.cs.ualberta.ca/~davidson/> | 
+-----------------------------------------------------------------------+