Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userCombining Deep Reinforcement Learning and Search for Imperfect-Information Games
93
0
0.0
(
0
)
Added by
Noam Brown
Publication date
2020
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
The combination of deep reinforcement learning and search at both training and test time is a powerful paradigm that has led to a number of successes in single-agent settings and perfect-information games, best exemplified by AlphaZero. However, prior algorithms of this form cannot cope with imperfect-information games. This paper presents ReBeL, a general framework for self-play reinforcement learning and search that provably converges to a Nash equilibrium in any two-player zero-sum game. In the simpler setting of perfect-information games, ReBeL reduces to an algorithm similar to AlphaZero. Results in two different imperfect-information games show ReBeL converges to an approximate Nash equilibrium. We also show ReBeL achieves superhuman performance in heads-up no-limit Texas holdem poker, while using far less domain knowledge than any prior poker AI.
rate research
Read More
Search has played a fundamental role in computer game research since the very beginning. And while online search has been commonly used in perfect information games such as Chess and Go, online search methods for imperfect information games have only been introduced relatively recently. This paper addresses the question of what is a sound online algorithm in an imperfect information setting of two-player zero-sum games. We argue that the~fixed-strategy~definitions of exploitability and $epsilon$-Nash equilibria are ill-suited to measure an online algorithms worst-case performance. We thus formalize $epsilon$-soundness, a concept that connects the worst-case performance of an online algorithm to the performance of an $epsilon$-Nash equilibrium. As $epsilon$-soundness can be difficult to compute in general, we introduce a consistency framework -- a hierarchy that connects an online algorithms behavior to a Nash equilibrium. These multiple levels of consistency describe in what sense an online algorithm plays just like a fixed Nash equilibrium. These notions further illustrate the difference between perfect and imperfect information settings, as the same consistency guarantees have different worst-case online performance in perfect and imperfect information games. The definitions of soundness and the consistency hierarchy finally provide appropriate tools to analyze online algorithms in repeated imperfect information games. We thus inspect some of the previous online algorithms in a new light, bringing new insights into their worst-case performance guarantees.
Counterfactual Regret Minimization (CFR) is an efficient no-regret learning algorithm for decision problems modeled as extensive games. CFRs regret bounds depend on the requirement of perfect recall: players always remember information that was revealed to them and the order in which it was revealed. In games without perfect recall, however, CFRs guarantees do not apply. In this paper, we present the first regret bound for CFR when applied to a general class of games with imperfect recall. In addition, we show that CFR applied to any abstraction belonging to our general class results in a regret bound not just for the abstract game, but for the full game as well. We verify our theory and show how imperfect recall can be used to trade a small increase in regret for a significant reduction in memory in three domains: die-roll poker, phantom tic-tac-toe, and Bluff.
Large-scale screening for potential threats with limited resources and capacity for screening is a problem of interest at airports, seaports, and other ports of entry. Adversaries can observe screening procedures and arrive at a time when there will be gaps in screening due to limited resource capacities. To capture this game between ports and adversaries, this problem has been previously represented as a Stackelberg game, referred to as a Threat Screening Game (TSG). Given the significant complexity associated with solving TSGs and uncertainty in arrivals of customers, existing work has assumed that screenees arrive and are allocated security resources at the beginning of the time window. In practice, screenees such as airport passengers arrive in bursts correlated with flight time and are not bound by fixed time windows. To address this, we propose an online threat screening model in which screening strategy is determined adaptively as a passenger arrives while satisfying a hard bound on acceptable risk of not screening a threat. To solve the online problem with a hard bound on risk, we formulate it as a Reinforcement Learning (RL) problem with constraints on the action space (hard bound on risk). We provide a novel way to efficiently enforce linear inequality constraints on the action output in Deep Reinforcement Learning. We show that our solution allows us to significantly reduce screenee wait time while guaranteeing a bound on risk.
Computing a Nash equilibrium (NE) is a central task in computer science. An NE is a particularly appropriate solution concept for two-agent settings because coalitional deviations are not an issue. However, even in this case, finding an NE is PPAD-complete. In this paper, we combine path following algorithms with local search techniques to design new algorithms for finding exact and approximate NEs. We show that our algorithms largely outperform the state of the art and that almost all the known benchmark game classes are easily solvable or approximable (except for the GAMUT CovariantGameRand class).
It is known that there are uncoupled learning heuristics leading to Nash equilibrium in all finite games. Why should players use such learning heuristics and where could they come from? We show that there is no uncoupled learning heuristic leading to Nash equilibrium in all finite games that a player has an incentive to adopt, that would be evolutionary stable or that could learn itself. Rather, a player has an incentive to strategically teach such a learning opponent in order secure at least the Stackelberg leader payoff. The impossibility result remains intact when restricted to the classes of generic games, two-player games, potential games, games with strategic complements or 2x2 games, in which learning is known to be nice. More generally, it also applies to uncoupled learning heuristics leading to correlated equilibria, rationalizable outcomes, iterated admissible outcomes, or minimal curb sets. A possibility result restricted to strategically trivial games fails if some generic games outside this class are considered as well.
suggested questions
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
