Während Libratus von Grund auf neu geschrieben wurde, ist es der nominelle Nachfolger von Claudico. Wie sein Vorgänger ist sein Name ein. Our goal was to replicate Libratus from a article published in Science titled Superhuman AI for heads-up no-limit poker: Libratus beats top professionals. According to one of Libratus' creators, Professor Tuomas Sandholm, Libratus does not have a fixed built-in strategy, but an algorithm that.
Künstliche Intelligenz: Poker-KI Libratus kennt kein Deep Learning, ist aber ein MultitalentPoker-Software Libratus "Hätte die Maschine ein Persönlichkeitsprofil, dann Gangster". Eine künstliche Intelligenz hat erfolgreicher gepokert. Die "Brains Vs. Artificial Intelligence: Upping the Ante" Challenge im Rivers Casino in Pittsburgh ist beendet. Poker-Bot Libratus hat sich nach. Our goal was to replicate Libratus from a article published in Science titled Superhuman AI for heads-up no-limit poker: Libratus beats top professionals.
Libratus Navigation menu VideoHow AI beat the best poker players in the world - Engadget R+D
Importantly, the Nash equilibria of zero-sum games are computationally tractable and are guaranteed to have the same unique value.
We define the maxmin value for Player 1 to be the maximum payoff that Player 1 can guarantee regardless of what action Player 2 chooses:. The minmax theorem states that minmax and maxmin are equal for a zero-sum game allowing for mixed strategies and that Nash equilibria consist of both players playing maxmin strategies.
As an important corollary, the Nash equilibrium of a zero-sum game is the optimal strategy. Crucially, the minmax strategies can be obtained by solving a linear program in only polynomial time.
While many simple games are normal form games, more complex games like tic-tac-toe, poker, and chess are not. In normal form games, two players each take one action simultaneously.
In contrast, games like poker are usually studied as extensive form games , a more general formalism where multiple actions take place one after another.
See Figure 1 for an example. All the possible games states are specified in the game tree. The good news about extensive form games is that they reduce to normal form games mathematically.
Since poker is a zero-sum extensive form game, it satisfies the minmax theorem and can be solved in polynomial time.
However, as the tree illustrates, the state space grows quickly as the game goes on. Even worse, while zero-sum games can be solved efficiently, a naive approach to extensive games is polynomial in the number of pure strategies and this number grows exponentially with the size of game tree.
Thus, finding an efficient representation of an extensive form game is a big challenge for game-playing agents. AlphaGo  famously used neural networks to represent the outcome of a subtree of Go.
While Go and poker are both extensive form games, the key difference between the two is that Go is a perfect information game, while poker is an imperfect information game.
In poker however, the state of the game depends on how the cards are dealt, and only some of the relevant cards are observed by every player.
To illustrate the difference, we look at Figure 2, a simplified game tree for poker. Note that players do not have perfect information and cannot see what cards have been dealt to the other player.
Let's suppose that Player 1 decides to bet. Player 2 sees the bet but does not know what cards player 1 has. In the game tree, this is denoted by the information set , or the dashed line between the two states.
An information set is a collection of game states that a player cannot distinguish between when making decisions, so by definition a player must have the same strategy among states within each information set.
Thus, imperfect information makes a crucial difference in the decision-making process. To decide their next action, player 2 needs to evaluate the possibility of all possible underlying states which means all possible hands of player 1.
Because the player 1 is making decisions as well, if player 2 changes strategy, player 1 may change as well, and player 2 needs to update their beliefs about what player 1 would do.
Heads up means that there are only two players playing against each other, making the game a two-player zero sum game.
No-limit means that there are no restrictions on the bets you are allowed to make, meaning that the number of possible actions is enormous.
In contrast, limit poker forces players to bet in fixed increments and was solved in . Nevertheless, it is quite costly and wasteful to construct a new betting strategy for a single-dollar difference in the bet.
Libratus abstracts the game state by grouping the bets and other similar actions using an abstraction called a blueprint. In a blueprint, similar bets are be treated as the same and so are similar card combinations e.
Ace and 6 vs. Ace and 5. The blueprint is orders of magnitude smaller than the possible number of states in a game. Libratus solves the blueprint using counterfactual regret minimization CFR , an iterative, linear time algorithm that solves for Nash equilibria in extensive form games.
Libratus uses a Monte Carlo-based variant that samples the game tree to get an approximate return for the subgame rather than enumerating every leaf node of the game tree.
It expands the game tree in real time and solves that subgame, going off the blueprint if the search finds a better action.
Solving the subgame is more difficult than it may appear at first since different subtrees in the game state are not independent in an imperfect information game, preventing the subgame from being solved in isolation.
This decouples the problem and allows one to compute a best strategy for the subgame independently. In short, this ensures that for any possible situation, the opponent is no better-off reaching the subgame after the new strategy is computed.
Thus, it is guaranteed that the new strategy is no worse than the current strategy. This approach, if implemented naively, while indeed "safe", turns out to be too conservative and prevents the agent from finding better strategies.
The new method  is able to find better strategies and won the best paper award of NIPS In addition, while its human opponents are resting, Libratus looks for the most frequent off-blueprint actions and computes full solutions.
Thus, as the game goes on, it becomes harder to exploit Libratus for only solving an approximate version of the game. While poker is still just a game, the accomplishments of Libratus cannot be understated.
Bluffing, negotiation, and game theory used to be well out of reach for artificial agents, but we may soon find AI being used for many real-life scenarios like setting prices or negotiating wages.
Soon it may no longer be just humans at the bargaining table. Correction: A previous version of this article incorrectly stated that there is a unique Nash equilibrium for any zero sum game.
The statement has been corrected to say that any Nash equilibria will have the same value. Thanks to Noam Brown for bringing this to our attention.
Citation For attribution in academic contexts or books, please cite this work as. If you enjoyed this piece and want to hear more, subscribe to the Gradient and follow us on Twitter.
Brown, Noam, and Tuomas Sandholm. Like its predecessor, its name is a Latin expression and means 'balanced'. Libratus was built with more than 15 million core hours of computation as compared to million for Claudico.
The computations were carried out on the new 'Bridges' supercomputer at the Pittsburgh Supercomputing Center. According to one of Libratus' creators, Professor Tuomas Sandholm, Libratus does not have a fixed built-in strategy, but an algorithm that computes the strategy.
Their new method gets rid of the prior de facto standard in Poker programming, called "action mapping". As Libratus plays only against one other human or computer player, the special 'heads up' rules for two-player Texas hold 'em are enforced.
To manage the extra volume, the duration of the tournament was increased from 13 to 20 days. The four players were grouped into two subteams of two players each.
One of the subteams was playing in the open, while the other subteam was located in a separate room nicknamed 'The Dungeon' where no mobile phones or other external communications were allowed.
The Dungeon subteam got the same sequence of cards as was being dealt in the open, except that the sides were switched: The Dungeon humans got the cards that the AI got in the open and vice versa.
This setup was intended to nullify the effect of card luck. As written in the tournament rules in advance, the AI itself did not receive prize money even though it won the tournament against the human team.
During the tournament, Libratus was competing against the players during the days. Overnight it was perfecting its strategy on its own by analysing the prior gameplay and results of the day, particularly its losses.
Therefore, it was able to continuously straighten out the imperfections that the human team had discovered in their extensive analysis, resulting in a permanent arms race between the humans and Libratus.
It used another 4 million core hours on the Bridges supercomputer for the competition's purposes. Libratus had been leading against the human players from day one of the tournament.
I felt like I was playing against someone who was cheating, like it could see my cards. It was just that good. This is considered an exceptionally high winrate in poker and is highly statistically significant.