People have been trying to design algorithms to solve different games for a long time. These algorithms already exist for games such as checkers and chess. Basically, these algorithms look at every possible move and figure out the perfect strategy to beat the opponent. However, certain games have an unpredictability factor. These are the so called "imperfect-information games," because certain information is hidden from each player. One of these games is poker. In order for computers to solve these games, they must handle the complication of not knowing certain information, which requires a lot of computer memory and power. But Neil Burch, a Ph.D. student in Computer Science, has successfully solved a two player version of Texas Hold'em by using an algorithm that he named CFR+. This is an upgrade to previous CFR algorithms that uses fewer, but bigger steps, essentially reducing the memory required and improving the program's efficiency.
More information can be found here: http://spectrum.ieee.org/tech-talk/computing/software/computers-conquer-texas-holdem-poker-for-first-time
Can you explain how he developed the algorithm? Also, why are people focusing on making algorithms to solve games?
ReplyDeleteI'm honestly surprised this hasn't been done earlier. After all, you can play card games online. What is the difference between this and online card games?
ReplyDeleteWhen they say "solved" they mean that they found a sure-fire way to win. Those online card games are either between users, playing online, or the computer just deals with its own side.
Delete