Print 29 comment(s) - last by oTAL.. on Jun 20 at 1:06 PM

Phil "The Unabomber" Laak
Move over Deep Blue, Polaris is here to take on Texas Hold 'Em poker

One of the best known matches between man and machine at a game is the IBM Deep Blue versus Garry Kasparov chess series. Now researchers from the University of Alberta, Canada, are putting computers up against the pros of Texas Hold 'Em Poker.

At a tournament worth $50,000 to be held July 23 - 24  at the Association for the Advancement of Artificial Intelligence's annual conference in Vancouver, B.C., poker pros Phil "The Unabomber" Laak and Ali Eslami will go up against Polaris, the reigning world champion computer-poker program.

"This is a world first and, I hope, the beginning of something that will grow and become an annual event," said Jonathan Schaeffer, a team leader of the Polaris program.

Schaeffer believes that making a winning program at poker is very different, and perhaps more difficult, challenge than the one IBM faced with Big Blue. "The difference is that chess is a game of perfect knowledge, meaning there is nothing hidden from the players. In poker you can't see your opponent's hand, and you don't know what cards will be dealt. This makes poker a much harder challenge for computer scientists from an artificial intelligence perspective," Schaeffer said.

The competition will feature four matches between Polaris and the two poker playing professionals. In each match, Laak and Eslami will play simultaneously against Polaris in separate rooms. At the end of each match, the players will combine their chip totals and compare them against Polaris' combined total.

Former World Series of Poker champion and host of the Mojo TV program I Bet You, Phil Laak is taking the challenge against Polaris as no common video game. "I am going to get in the zone. They'll be getting my best game," he said. "I am treating this with all seriousness - it would be very embarrassing if Prince Ali broke even but the Unabomber got whacked. If we have an amazing showing, I want the headline to be a testament to the guys at the U of A - that they designed a program that can play on terms with phenomenal players."

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

By sprockkets on 6/12/2007 5:11:10 PM , Rating: 3
So, uh, will they have like a a way of showing whether the computer is bluffing, like you know, hard drive light keeps flashing every other second or maybe a mouse icon on the screen keeps fidgeting haha

RE: ok
By JarredWalton on 6/12/2007 5:19:56 PM , Rating: 2
I thought the point of a tournament was to play until one player busted the other? What's with the four matches, combining chip totals, etc. junk? I'd like to see eight pros go up against the computer in several tourneys and see how it does in placing. Maybe with a neural net or something it could learn player betting styles?

Will it throw artificial tantrums to try and bluff or confuse opponents? I can see it now:

Phil sweats for a while, trying to decide whether to call or not. Announcers try to build up tension. Phil decides to raise.... The computer responds .5 nanoseconds later by calling. Or folding. Or whatever.

Poker is as much a game of learning to read others as it is knowing your odds of having a winning hand. I'm not sure if the computer is going to be seriously handicapped because it's an algorithm, or if the inability to get a read will throw off the pros.

RE: ok
By aos007 on 6/12/2007 5:34:09 PM , Rating: 2
I'm sure in a few years they will be adding cameras and software to analyze human player's facial expressions as one of the components of the computer player's program. Isn't technology of that type being already developed?

RE: ok
By sotti on 6/12/2007 5:34:19 PM , Rating: 3
All the big money is in the never ending cash games that go on, not the tournaments.

People play online and are very succeful where there are no tells.

As for the computer it can measure and track a very large number of hands and have better recall and anaylsis than a human opponent, as well as a truly randomizing factor in it's own play.

Poker is a game of anaylsing patterns, adjusting your play and manipulating odds.

RE: ok
By stromgald on 6/12/2007 7:29:31 PM , Rating: 2
Online poker has tells also. You can analyze the speed at which a player makes his moves to try to guess at what he's trying to do. Factor in previous knowledge of that opponent and knowledge about human nature and you have some bluffs and tells.

If a computer doesn't have any tells and bluffs and doesn't count cards, then it just has to compute the odds and do what is theoretically the best. If that's the case, there doesn't seem to be much of a point. The human might win in a handful of games, but in the long run, the computer should come out on top since a human doesn't compute all the odds.

RE: ok
By exanimas on 6/12/2007 10:56:57 PM , Rating: 1
Not to nit-pick but you have to factor in a few other things. Just a for instance but you notice that say, PokerMan13987, takes an extra long time to bet when he is bluffing. You notice this pattern and bet/play according to this. PokerMan13987 decides to get up and get a drink, while up its his turn, he gets back just in time to make his bet with pocket aces and 2 more on the flop, you lose.

That scenario was a bit dramatic, but my short point is that judging by timing alone shouldn't be considered a tell. Also, playing with the odds gives you better theoretical chance, but I've seen people win by getting the 1 card in the deck they need on the river. Unless the can give the computer e-balls, it won't take the risks that end up winning (some) games.

RE: ok
By Scorpion on 6/12/2007 6:32:12 PM , Rating: 2
From what little knowledge I have, this computer will just be computing the probabilities of outcomes and take action according to that. In that respect a computer is precise, and has a clear advantage over the human.

How will new hands be done? Will they play through with a single deck, or will a new simulated deck be constructed on every play? Because a computer can easily count cards and compute the probability of drawing a certain card.

Now if they were to build a camera vision system and have the computer adaptively try to recognize the humans tells then that would be interesting, but it seems that the guy has no way of getting tells from the computer, unless ones are artificially introduced.

This really just seems rather silly and a ploy to draw some attention. Doh.

RE: ok
By BMFPitt on 6/13/2007 12:11:49 AM , Rating: 3
From what little knowledge I have, this computer will just be computing the probabilities of outcomes and take action according to that. In that respect a computer is precise, and has a clear advantage over the human.
If the computer is precise, it is predictable, and therefore very easy to beat. The AI has to be able to vary it's play to succeed. And every player at the table will know the odds, too. It's a relatively small set that most have memorized (many, many hands are equivalent as far as computing odds.)

How will new hands be done? Will they play through with a single deck, or will a new simulated deck be constructed on every play? Because a computer can easily count cards and compute the probability of drawing a certain card.
Well considering the article says "poker" and not "some made-up game where the deck doesn't get shuffled between hands."

RE: ok
By TrogdorJW on 6/13/2007 12:37:37 AM , Rating: 2
What's with the talk about counting cards? It's POKER, not blackjack. Man, don't any of you people play? You reshuffle on every hand. One deck, with at most eight players, and each gets two cards plus the five community cards. Sure you could play two hands without a reshuffle, but that's not the way poker is played. Imagine someone getting four aces in a hand; have you ever seen a poker tourney where they then say, "all the aces were in play last hand, so there's none this time"?

If the computer just plays the odds, the winner is going to be as much luck as anything. Both players will know the odds of each hand winning, and if the computer can't read the human (or doesn't) and the human can't read the computer, it's going to come down to chance. My bet is that the computer is programmed to actually act intelligent and not just play odds, so timing will certainly be a factor as well as previous bet history. Otherwise, this would be a trivial task to program and there would be no need of a super computer.

RE: ok
By Moishe on 6/13/2007 8:08:14 AM , Rating: 2
the good thing (and bad thing) about poker vs a computer is that you cannot read the computer and get a sense of it's position. Good thing is that it goes both ways. The computer cannot read you.

It's cool because the game is so dependent on some intangible human ability to read the body language, etc of another human. playing against a computer removes that and it becomes more about making sure you're playing it within safe margins.

I already know the outcome...
By Souka on 6/12/2007 7:03:25 PM , Rating: 2
Come on... we all know the outcome... human will win.

Think about it... Lt. Commander Data always lost to Commander Riker, and I'm sure we'll all agree that Data is more advanced than the computer in this tournament....

RE: I already know the outcome...
By Brandon Hill on 6/12/2007 7:20:20 PM , Rating: 6
Yeah, but Data managed to score with Tasha Yar. I don't think that Riker achieved that feat :)

RE: I already know the outcome...
By MarkHark on 6/12/2007 7:43:54 PM , Rating: 2
That was only 'cause she was drunk :)
AND she later said that it "never happened"

RE: I already know the outcome...
By KristopherKubicki on 6/13/2007 6:53:16 AM , Rating: 4
OK, are you guys just trying to figure out who is the bigger nerd? I thought posting on DailyTech would be proof enough that you're booth in the top 0.01% of Nerdom.

And besides, a little polywater and Tasha gave it up to just about anyone. Figures the robot didn't pass her up! 8D

RE: I already know the outcome...
By Moishe on 6/13/2007 8:26:37 AM , Rating: 2
LOL.... true. Tasha was the "babe" of the show at the time. Only Troi beat her out as the long term ship's "counselor". There was nothing she couldn't fix with a little couch time.

By theapparition on 6/13/2007 8:39:06 AM , Rating: 2
Souka = Star Trek loser nerd
Brandon = Star Trek loser nerd
MarkHark = Star Trek loser nerd
Kris = Star Trek loser nerd
Moishe = Star Trek loser nerd

me = Star Trek loser nerd (since I knew what you were talking about and laughed my ass off)

Kevin Costner
By Brandon Hill on 6/12/2007 5:00:58 PM , Rating: 3
That guy looks like a young, street punk'd version of Kevin Costner.

RE: Kevin Costner
By henrikfm on 6/12/2007 5:08:36 PM , Rating: 2
:D You're right!

RE: Kevin Costner
By ancient46 on 6/13/2007 9:22:43 AM , Rating: 2
Phil's Girlfriend and poker partner = Jennifer Tilly

This isn't like chess
By Nik00117 on 6/13/2007 7:54:13 AM , Rating: 3
One in poker can win on all luck alone. Trust me I am a experienced poker player and have lost several times to completely newbs at the game. It could turn out to be a simple game of luck.

Although I don't see the difficulty in the AI coding, I mean just gotta tell the PC what good cards are and what bad cards etc. Then it can do its odds and etc and decide to check, bet, raise, match or fold etc.

RE: This isn't like chess
By KentState on 6/13/2007 2:01:49 PM , Rating: 2
For an "experienced" poker player, you really missed things that AI will need to code for. For example, if a player raises before the flop, that has to be considered. Betting tendencies are just as important as the cards in hand if not more important. The computer will have to be able to bet well enough to not allow the humans to know what type of hand it has.

Plus, "experienced" poker players know the odds of the cards available to them.

RE: This isn't like chess
By oTAL on 6/20/2007 1:06:54 PM , Rating: 2
Well, you are right about poker (and you probably know a lot more about it since I only play with friends) but you are wrong about the AI (and I'm pretty sure I know a lot more about that).

As you said, poker isn't like chess. There are two main things that make it very different to program:
1 - You don't know everything. The opponents cards are a mystery.
2 - Neither of you knows which cards will be placed on the table.

I distinguish both these factors since they must be handled differently.

A program can easily learn statistics and know the odds of any card being drawn and which hand is most probable to win.

On the other hand, it is a lot harder to effectively learn how a certain opponent reacts to a good or a bad hand or how he adapts his play to the amount of chips each player has and the size of the pot on the table. It is possible. It can even learn how to use bluffs and when to call or fold. But this is extremely hard to program and is compounded by the fact that you must "reward" the computer when it looses if the decision he made had the best probability of success (and "punish" him, even when he wins against the odds). Furthermore, you can't implement a decent reinforcement learning algorithm if you don't see your opponent's cards when any of you fold (this is a BIG deal since this will make it much harder for you to understand your opponent and evaluate your decisions). In training the computer should probably cheat and see the opponents cards when anyone folds.
Plus, even if they make the perfect program, but they make it deterministic (= same situation->same output), the opponents will find it easier to predict on certain situations. Obviously this isn't simple since the opponent does not know your cards, but it can be exploited in some situations. That means a small amount of randomness (or intolerance towards repetition) should be introduced on certain decisions.

This is a very difficult problem and this first time around, I would bet on the humans.
By next year, if the team is committed to the problem, I would bet on the computer. Why? Because the problem here is finding and fine tuning the best algorithms. The problem with chess is mostly lack of computer power for the needed calculations. You need algorithms that provide the best moves without taking years to compute. In poker you don't require that much computing power; only good algorithms, a very good implementation of reinforcement learning, and lots of time to play against the program (and to make it play against other computers).
The current situation is that a chess game can be lost because computers aren't fast enough, while a poker game will be lost because the algorithms weren't good enough (or simply bad luck since there's randomness involved). The first problem will not be easy to solve since you need massive improvements in computing power. The second can be solved by a single brilliant person.

Obviously, when you're at the tournament you will also require a bit of luck.

Bluffs and Tells
By techfuzz on 6/12/2007 5:52:50 PM , Rating: 2
The last time I checked, poker involved bluffs and tells which coincidentally computers are not capable of. So essentially we're going to have a computer play the odds and the person play poker. Should that be the case, my predication: computer wins every game.

RE: Bluffs and Tells
By FeralMisanthrope on 6/13/2007 11:45:27 AM , Rating: 2
I think you are underestimating the sophistication of this software. There is no reason why a computer can't bluff. In fact, bluffing is essential to any advantage the computer may have since tells aren't really an issue. Simply "playing the odds" would be a terrible strategy, betraying the strength (or weakness) of the computer's hand and making it impossible for the computer to call an opponent's bluff.

While you are correct that there are no visual tells involved, this doesn't give either opponent an advantage. It will be just like playing poker online. In fact, many people profitably employ poker-bots in online poker rooms despite the questionable legality.

They're taking over!
By MatthewAC on 6/12/2007 7:30:28 PM , Rating: 2
Computers are starting to beat us at almost all mental tasks.

Sarcasm aside, I think this is a interesting project but for pure publicity, without the publicity this would be useless.

RE: They're taking over!
By KristopherKubicki on 6/13/2007 7:11:00 AM , Rating: 2
If you think about it - IBM's Deep Blue against Garry Kasparov was the same thing ... it's just that Texas Holdem is more popular now.

By jmunjr on 6/12/2007 6:47:59 PM , Rating: 2
If a computer is equipped with a camera and other sensors and software to detect "tells" then it would be a HUGE advantage for the computer. Computers won't give any tells, but will be taught how to detect them? That's an unfair advantage.

Without cameras, going head to head a computer might fair well, but in a large group it would get dominated imo.

By rbowling on 6/12/2007 7:24:50 PM , Rating: 2
Physical tells in poker are severely overrated. That is why so many people can come from the internet and do okay in live tournaments. In-play tells are very real though. People develop a style of play and respond to similar situations similarly in the way they raise and call. The computer should be able to assimilate that information and use it.

I think the programmers should be able to beat the computer pretty easily, but a random poker player who doesn't understand how the computer is making decisions will likely get hosed.

Computer duels?
By MonkeyPaw on 6/12/2007 7:39:15 PM , Rating: 2
I'd be more satisfied if Polaris "beat" that snotty Mac guy from those annoying Apple commercials.

At least "PC" offers "Hold'em" as an Vista Ultimate extra. :p

“And I don't know why [Apple is] acting like it’s superior. I don't even get it. What are they trying to say?” -- Bill Gates on the Mac ads

Copyright 2016 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki