Computer beats Pro Poker Players

Discuss the latest news and rumors in the magic world.
User avatar
erdnasephile
Posts: 4770
Joined: January 17th, 2008, 12:00 pm

Computer beats Pro Poker Players

Postby erdnasephile » March 5th, 2017, 3:08 pm

https://www.msn.com/en-us/news/technolo ... ar-AAnKcBD

(If you have Al Stanger's trick, there are interesting patter possibilities suggested by this current news story)

User avatar
erdnasephile
Posts: 4770
Joined: January 17th, 2008, 12:00 pm

Re: Computer beats Pro Poker Players

Postby erdnasephile » March 5th, 2017, 3:41 pm

Here's a link to actual research article: http://science.sciencemag.org/content/e ... m6960.full

Roger M.
Posts: 1598
Joined: January 17th, 2008, 12:00 pm

Re: Computer beats Pro Poker Players

Postby Roger M. » March 5th, 2017, 4:05 pm

I wonder how the computer would fare in a "No Limit" game?

I suspect they choose "Limit Poker" for their demo for this reason.

User avatar
erdnasephile
Posts: 4770
Joined: January 17th, 2008, 12:00 pm

Re: Computer beats Pro Poker Players

Postby erdnasephile » March 6th, 2017, 8:44 am

I think in the study the authors stated the game the computer played was Heads-up no-limit Texas hold’em (HUNL).

Roger M.
Posts: 1598
Joined: January 17th, 2008, 12:00 pm

Re: Computer beats Pro Poker Players

Postby Roger M. » March 6th, 2017, 10:06 am

Ahhh, thanks for that, re-reading the articles they do indeed reference no-limit as the final test, after they had conquered limit. My misread.

My next question then, would be who were the "pros" that the computer beat?
No names were given, and player skills vary widely, with many who consider themselves "pros" - nothing of the sort.

If one of the players was Daniel Negreanu, I'm convinced ... but if it's a name I don't recognize, I'd need to read a lot more details about how the testing took place.

User avatar
erdnasephile
Posts: 4770
Joined: January 17th, 2008, 12:00 pm

Re: Computer beats Pro Poker Players

Postby erdnasephile » March 6th, 2017, 11:15 am

Fair point--the quality of competition is central to this study.

Like most research studies, the names of the subjects were not disclosed:

"We evaluated DeepStack by playing it against a pool of professional poker players recruited by the International Federation of Poker (36). Thirty-three players from 17 countries were recruited. Each was asked to complete a 3,000 game match over a period of four weeks between November 7th and December 12th, 2016. Cash incentives were given to the top three performers ($5,000, $2,500, and $1,250 CAD)."

The one name that got out in the news was Martin Sturc. He was apparently the only one of 11 players who completed the entire 3,000 game match who statistically tied the computer. (The other 10 lost by wide margins).

I know nothing about professional poker. Is Mr. Sturc a good player?

-------------------------

A rival artificial intelligence, Libratus, at Carnegie Mellon beat 4 pros in January: Dong Kim, Jimmy Chou, Daniel McAulay and Jason Les.

https://www.cmu.edu/news/stories/archiv ... -pros.html

------------------------
What I found fascinating is a statistic quoted in both articles:

"Head’s-Up No-Limit Texas Hold’em is an exceedingly complex game, with...[10 to the 160th power]...(the number 1 followed by 160 zeroes) information sets — each set being characterized by the path of play in the hand as perceived by the player whose turn it is. That’s vastly more information sets than the number of atoms in the universe."

User avatar
lybrary
Posts: 1169
Joined: March 31st, 2013, 4:59 pm
Contact:

Re: Computer beats Pro Poker Players

Postby lybrary » March 6th, 2017, 12:53 pm

The most fascinating aspect of this development is that computers starting to break into domains and beating humans which were previously firmly held in human hands. When I was at university, a good 20 years ago, AI was considered a big failure. Some folks dabbled with neural networks but what they could do with them was very limited. My supervisor, who was tightly connected to the semiconductor industry pretty much predicted what is happening today. He said in 2-3 decades computers will start to solve real problems previously considered too hard or too complex, because of continuing exponential growth (faster, smaller, cheaper) of semiconductors. Various recent achievements in the filed of "deep learning" are all driven by neural networks, now several orders of magnitude more complex than 20 years ago, which are possible due to ever improving integrated circuits built on semiconductor technology.
Lybrary.com Magic & Gambling
preserving magic one book at a time


Return to “Buzz”