Man vs. Machine: Poker Pros To Put AI to the Test Jan. 11
Poker professionals will compete against new artificial intelligence again developed by Carnegie Mellon University to see whether the odds favor man or machine in heads-up no-limit Texas Hold��em games.
Beginning Jan. 11 at Pittsburgh��s Rivers Casino, the ��Brains Vs. Artificial Intelligence: Upping the Ante�� event will host four poker pros �C Jason Les, Dong Kim, Daniel McAulay and Jimmy Chou �C who will then play 120,000 hands of the heads-up no-limit hold��em games over the course of 20 days for a share of a $200,000 prize.
The computer of choice, Libratus, has been created since last year��s competition when Claudico, the former Carnegie Mellon computer program collected less chips than three of the four players last year, playing out 80,000 hands, too few to determine a conclusion of whether man or machine outwitted the other with statistical significance.
The pros will pair up and play duplicate matches (with each pair receiving the same cards as the computer in each scenario) on the casino floor and in an isolated separate room. The increased number of days and the ��two-table�� play (playing two hands simultaneously) will increase the chance of reaching statistical significance. Play will begin at 1 a.m. and end around 7 p.m. each day in the Rivers�� Poker Room and will be open for public viewing.
This year��s goal is to set a new benchmark for artificial intelligence.
��Since the earliest days of AI research, beating top human players has been a powerful measure of progress in the field,�� said Tuomas Sandholm, the professor of computer science at Carnegie Mellon who created the bot with his Ph.D student Noam Brown, in a release from the university. ��That was achieved with chess in 1997, with Jeopardy! in 2009 and with the board game Go just last year. Poker poses a far more difficult challenge than these games, as it requires a machine to make extremely complicated decisions based on incomplete information while contending with bluffs, slow play and other ploys.��
With both players and the artificial intelligence making some adjustments to improve their play, it may be difficult to predict the outcome.
Poker poses a far more difficult challenge than these games, as it requires a machine to make extremely complicated decisions based on incomplete information.
This Libratus artificial intelligence poker bot was created from scratch with an algorithm developed to compute strategies for imperfect information games and to use the Pittsburgh Supercomputing Center��s Bridge supercomputer to calculate the potential winning strategy.
Carnegie Mellon used 15 million core hours of computation for Libratus versus the two to three million core hours for Claudico. Other improvements include some ��weird moves�� like limping (a favored strategy of the old bot which was exploited by the players) and new technology to achieve Nash equilibrium �C a strategy that neither player can benefit from changing strategy if the other player��s strategy remains the same.
Libratus, a Latin word meaning balanced and powerful, was chosen because of this incorporation of the new technology in recognition of Carnegie Mellon alumnus and Nobel laureate John Forbes Nash Jr. The bot will also use a method to find equilibrium faster, identifying hands that are not promising and ignoring these paths in the future, and use the Bridges computer to do live computations with a new endgame-solving approach and algorithm.
The complexities of heads-up (two player) no-limit hold��em due to the number of information sets (10160) and the perception of the path of play in the hand by the player��s whose turn it is forces the bot to make decisions without knowing all the cards in play and try to adjust for bluffing.
This sort of game has implications for other sectors that are characterized by incomplete and misleading information, like business, military, cybersecurity and medicine.
��Extending AI to real-world decision-making, where details are unknown and adversaries are actively revising their strategies, is fundamentally harder than games with perfect information or question-answering systems,�� said Nick Nystrom, senior director of research at the Pittsburgh Supercomputing Center, in the university's release. ��This is where it really gets interesting.��
In February 2016, a bot developed by Sandholm and Brown won both categories of heads-up no-limit hold��em in the Annual Computer Poker Competition. An announcement was made at the Association for the Advancement of Artificial Intelligence Conference in Arizona.
A computer poker group at the University of Alberta solved the game of heads-up limit hold��em near-optimally.
Brains Vs. AI is sponsored by GreatPoint Ventures, Avenue4Analytics,TNG Technology Consulting GmbH, the journal Artificial Intelligence,Intel and Optimized Markets, Inc. Carnegie Mellon��s School of Computer Science has partnered with Rivers Casino, the Pittsburgh Supercomputing Center (PSC) through a peer-reviewed XSEDE allocation and Sandholm��sElectronic Marketplaces Laboratory for this event.
Lead image courtesy of Carnegie Mellon University.
