Carnegie Mellon Computer Challenges World's Best Poker Players At Texas Hold 'Em
Carnegie Mellon University researchers have created a new computer program which will go face-to-interface against some of the top poker players in the world. The game of Texas Hold 'Em will be a no-limit game, meaning players, both human and electronic, can wager any amount of money.
Claudico was developed by CMU investigators from the institution's School of Computer Science. The program will play 20,000 hands each against four of the top Texas Hold 'Em players in the world - Dong Kim, Bjorn Li, Doug Polk and Jason Les.
Games once thought to be the purview of humans are being mastered by computers in recent years. Deep Blue won a game of chess against master player Garry Kasparov, and a computer named Watson took down a pair of human opponents in the game show Jeopardy.
The Brains vs. Artificial Intelligence competition will be held at Rivers Casino in Pittsburgh, beginning April 24 and lasting through May 8.
While the human players are competing in exchange for a share of $100,000 in prize money, the CMU developers have their minds set on proving their programming can beat some of the best human players on the planet. The prize money was donated by Microsoft Research and Rivers Casino.
"Poker is now a benchmark for artificial intelligence research, just as chess once was. It's a game of exceeding complexity that requires a machine to make decisions based on incomplete and often misleading information, thanks to bluffing, slow play, and other decoys. And to win, the machine has to out-smart its human opponents," Tuomas Sandholm, leader of the team at CMU that developed Claudico, said.
No-limit Texas Hold 'Em is usually considered to be a far more challenging game than the Limit version of the game, where bets are restricted. Play will consist of two 750-hand sessions each day over the two-week tournament, with one day of rest for the human players. The cards dealt to each player will also be matched between players over the various rounds - the players will be paired to play duplicate matches, as Player A will receive the same cards as the computer receives against Player B, and vice versa. This, combined with the massive number of hands to be played, is designed to ensure that victory depends on playing ability, rather than luck. Players will be isolated to keep them from revealing information about the hands they receive.
"I think it's a 50-50 proposition. I think there's a good chance we'll lose this thing. I imagine that the humans have an edge here. However, it is very difficult to determine an outcome with any sort of stability, as I do not know what I am going to be up against," said Polk, often considered the best player of the game. He has won more than $3.6 million over his career at live-action tournaments.
Development of the algorithms needed to decide on winning strategies could have implications for artificial intelligence in machines operated by medical staff, private businesses, and the military. The computer Claudico will play via a university computer but will be able to tap the power of Pittsburgh's Supercomputing Center's Blacklight supercomputer to help compute Claudico's strategy.
Deep Blue, Watson and now Claudico? I'll see your two computer players and raise you one.