CMU’s Second Poker Robot will go Head to Head with Humans
Carnegie Mellon University’s poker robot, Libratus, will be duelling
human opponents on January 11, in an attempt at a milestone: an AI machine winning an incomplete-information game.
Daniel McAuley, Jimmy Chou, Dong Kim and Jason Les are the
poker pros that will play against Libratus in a series of hands, over the
course of the 20-day tournament. The clash will take place at Rivers Casino, in
Pittsburgh and the prize for the winner is $200,000.
Libratus was created by Tuomas Sandholm, professor of
Computer Science at Carnegie Mellon, and Noam Brown, a doctoral student under his advisory. This is the second AI machine specialized for poker games created
at Carnegie Mellon. The first iteration, Claudico, lost to another group of
players, that include two of this year’s challengers.
The poker robot Libratus was created with an algorithm
designed to study strategies of this type of game and has been tested over 15 million hours. Other
games that AI have already conquered, Chess and Go, are perfect-information
games: each player knows exactly what moves have been made and what space is
left on the board to consider. Poker is an imperfect-information game, which
makes it far more challenging for artificial intelligence to master.
Read the Portuguese article at PokerPT.com (January 5, 2017).