We consider a scenario in which two reinforcement learning agents repeatedly play a matrix game against each other and update their parameters after each round. The agents' decision-making is transparent to each other, which allows each agent to predict how their opponent will play against them. To prevent an infinite regress of both agents recursively predicting each other indefinitely, each agent is required to give an opponent-independent response with some probability at least epsilon. Transparency also allows each agent to anticipate and shape the other agent's gradient step, i.e. to move to regions of parameter space in which the opponent's gradient points in a direction favourable to them. We study the resulting dynamics experimentally, using two algorithms from previous literature (LOLA and SOS) for opponent-aware learning. We find that the combination of mutually transparent decision-making and opponent-aware learning robustly leads to mutual cooperation in a single-shot prisoner's dilemma. In a game of chicken, in which both agents try to manoeuvre their opponent towards their preferred equilibrium, converging to a mutually beneficial outcome turns out to be much harder, and opponent-aware learning can even lead to worst-case outcomes for both agents. This highlights the need to develop opponent-aware learning algorithms that achieve acceptable outcomes in social dilemmas involving an equilibrium selection problem.
This paper proposes new, end-to-end deep reinforcement learning algorithms for learning two-player zero-sum Markov games. Different from prior efforts on training agents to beat a fixed set of opponents, our objective is to find the Nash equilibrium policies that are free from exploitation by even the adversarial opponents. We propose (a) Nash-DQN algorithm, which integrates the deep learning techniques from single DQN into the classic Nash Q-learning algorithm for solving tabular Markov games; (b) Nash-DQN-Exploiter algorithm, which additionally adopts an exploiter to guide the exploration of…
University of Oxford（牛津大学）；DeepMind（深度思维）；University College London（伦敦大学学院）
A growing number of learning methods are actually differentiable games whose players optimise multiple, interdependent objectives in parallel -- from GANs and intrinsic curiosity to multi-agent RL. Opponent shaping is a powerful approach to improve learning dynamics in these games, accounting for player influence on others' updates. Learning with Opponent-Learning Awareness (LOLA) is a recent algorithm that exploits this response and leads to cooperation in settings like the Iterated Prisoner's Dilemma. Although experimentally successful, we show that LOLA agents can exhibit 'arrogant' behavio…
Two player zero sum simultaneous action games are common in video games, financial markets, war, business competition, and many other settings. We first introduce the fundamental concepts of reinforcement learning in two player zero sum simultaneous action games and discuss the unique challenges this type of game poses. Then we introduce two novel agents that attempt to handle these challenges by using joint action Deep Q-Networks (DQN). The first agent, called the Best Response AgenT (BRAT), builds an explicit model of its opponent's policy using imitation learning, and then uses this model t…
Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所）；School of Artificial Intelligence, University of Chinese Academy of Sciences（中国科学院大学人工智能学院）；Queen Mary University of London（伦敦玛丽女王大学）
The Fighting Game AI Competition (FTGAIC) provides a challenging benchmark for 2-player video game AI. The challenge arises from the large action space, diverse styles of characters and abilities, and the real-time nature of the game. In this paper, we propose a novel algorithm that combines Rolling Horizon Evolution Algorithm (RHEA) with opponent model learning. The approach is readily applicable to any 2-player video game. In contrast to conventional RHEA, an opponent model is proposed and is optimized by supervised learning with cross-entropy and reinforcement learning with policy gradient …
Vrije Universiteit Brussel（布鲁塞尔自由大学）；Universiteit van Amsterdam（阿姆斯特丹大学）；National University of Ireland Galway（爱尔兰国立大学戈尔韦分校）；HU Univ. of Appl. Sci. Utrecht（乌得勒支HU应用科学大学）
Many real-world multi-agent interactions consider multiple distinct criteria, i.e. the payoffs are multi-objective in nature. However, the same multi-objective payoff vector may lead to different utilities for each participant. Therefore, it is essential for an agent to learn about the behaviour of other agents in the system. In this work, we present the first study of the effects of such opponent modelling on multi-objective multi-agent interactions with non-linear utilities. Specifically, we consider two-player multi-objective normal form games with non-linear utility functions under the sca…
Boston University（波士顿大学）；Center for Brains, Minds and Machines, MIT（麻省理工学院脑、心智与机器中心）；CMSA, Harvard University（哈佛大学数学与应用中心）
Reinforcement learning has shown much success in games such as chess, backgammon and Go. However, in most of these games, agents have full knowledge of the environment at all times. In this paper, we describe a deep learning model in which agents successfully adapt to different classes of opponents and learn the optimal counter-strategy using reinforcement learning in a game under partial observability. We apply our model to FlipIt, a two-player security game in which both players, the attacker and the defender, compete for ownership of a shared resource and only receive information on the cur…
In this paper, we introduce a skill-balancing mechanism for adversarial non-player characters (NPCs), called Skilled Experience Catalogue (SEC). The objective of this mechanism is to approximately match the skill level of an NPC to an opponent in real-time. We test the technique in the context of a First-Person Shooter (FPS) game. Specifically, the technique adjusts a reinforcement learning NPC's proficiency with a weapon based on its current performance against an opponent. Firstly, a catalogue of experience, in the form of stored learning policies, is built up by playing a series of training…
Bilkent University（比尔肯特大学）；Cornell University（康奈尔大学）；Massachusetts Institute of Technology（麻省理工学院）
We present a novel variant of fictitious play dynamics combining classical fictitious play with Q-learning for stochastic games and analyze its convergence properties in two-player zero-sum stochastic games. Our dynamics involves players forming beliefs on the opponent strategy and their own continuation payoff (Q-function), and playing a greedy best response by using the estimated continuation payoffs. Players update their beliefs from observations of opponent actions. A key property of the learning dynamics is that update of the beliefs on Q-functions occurs at a slower timescale than update…
This technical report documents the winner of the Computational Intelligence in Games(CIG) 2018 Hanabi competition. We introduce Re-determinizing IS-MCTS, a novel extension of Information Set Monte Carlo Tree Search (IS-MCTS) that prevents a leakage of hidden information into opponent models that can occur in IS-MCTS, and is particularly severe in Hanabi. Re-determinizing IS-MCTS scores higher in Hanabi for 2-4 players than previously published work at the time of the competition. Given the 40ms competition time limit per move we use a learned evaluation function to estimate leaf node values a…
Singapore Management University（新加坡管理大学）；University of Oregon（俄勒冈大学）
Recent research on vulnerabilities of deep reinforcement learning (RL) has shown that adversarial policies adopted by an adversary agent can influence a target RL agent (victim agent) to perform poorly in a multi-agent environment. In existing studies, adversarial policies are directly trained based on experiences of interacting with the victim agent. There is a key shortcoming of this approach; knowledge derived from historical interactions may not be properly generalized to unexplored policy regions of the victim agent, making the trained adversarial policy significantly less effective. In t…
Efficient Adaptation in Mixed-Motive Environments via Hierarchical Opponent Modeling and Planning
Institute for Artificial Intelligence, Peking University（北京大学人工智能研究院）；State Key Laboratory of General Artificial Intelligence, BIGAI（北京通用人工智能研究院通用人工智能国家重点实验室）；University of California, Los Angeles（加利福尼亚大学洛杉矶分校）；Tsinghua University（清华大学）
Coordinated Science Laboratory（协同科学实验室）；University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）
Making decisions in the presence of a strategic opponent requires one to take into account the opponent's ability to actively mask its intended objective. To describe such strategic situations, we introduce the non-cooperative inverse reinforcement learning (N-CIRL) formalism. The N-CIRL formalism consists of two agents with completely misaligned objectives, where only one of the agents knows the true objective function. Formally, we model the N-CIRL formalism as a zero-sum Markov game with one-sided incomplete information. Through interacting with the more informed player, the less informed p…
We propose the first model-free algorithm that achieves low regret performance for decentralized learning in two-player zero-sum tabular stochastic games with infinite-horizon average-reward objective. In decentralized learning, the learning agent controls only one player and tries to achieve low regret performances against an arbitrary opponent. This contrasts with centralized learning where the agent tries to approximate the Nash equilibrium by controlling both players. In our infinite-horizon undiscounted setting, additional structure assumptions is needed to provide good behaviors of learn…
StarCraft, one of the most popular real-time strategy games, is a compelling environment for artificial intelligence research for both micro-level unit control and macro-level strategic decision making. In this study, we address an eminent problem concerning macro-level decision making, known as the 'fog-of-war', which rises naturally from the fact that information regarding the opponent's state is always provided in the incomplete form. For intelligent agents to play like human players, it is obvious that making accurate predictions of the opponent's status under incomplete information will i…
We study a version of the classical zero-sum matrix game with unknown payoff matrix and bandit feedback, where the players only observe each others actions and a noisy payoff. This generalizes the usual matrix game, where the payoff matrix is known to the players. Despite numerous applications, this problem has received relatively little attention. Although adversarial bandit algorithms achieve low regret, they do not exploit the matrix structure and perform poorly relative to the new algorithms. The main contributions are regret analyses of variants of UCB and K-learning that hold for any opp…
In current state-of-the-art commercial first person shooter games, computer controlled bots, also known as non player characters, can often be easily distinguishable from those controlled by humans. Tell-tale signs such as failed navigation, "sixth sense" knowledge of human players' whereabouts and deterministic, scripted behaviors are some of the causes of this. We propose, however, that one of the biggest indicators of non humanlike behavior in these games can be found in the weapon shooting capability of the bot. Consistently perfect accuracy and "locking on" to opponents in their visual fi…
We present the design of a competitive artificial intelligence for Scopone, a popular Italian card game. We compare rule-based players using the most established strategies (one for beginners and two for advanced players) against players using Monte Carlo Tree Search (MCTS) and Information Set Monte Carlo Tree Search (ISMCTS) with different reward functions and simulation strategies. MCTS requires complete information about the game state and thus implements a cheating player while ISMCTS can deal with incomplete information and thus implements a fair player. Our results show that, as expected…
Department of Computer Science, University College London（计算机科学系，伦敦大学学院）；Centre for Artificial Intelligence, University College London（人工智能中心，伦敦大学学院）；Department of Computer Science, University of Bologna（计算机科学系，博洛尼亚大学）
Departamento de Engenharia Informática, Faculdade de Engenharia da Universidade do Porto（波尔图大学工程学院信息工程系）；Laboratório de Inteligência Artificial e de Ciência de Computadores, Universidade do Porto（波尔图大学人工智能与计算机科学实验室）
The development of competitive artificial Poker playing agents has proven to be a challenge, because agents must deal with unreliable information and deception which make it essential to model the opponents in order to achieve good results. This paper presents a methodology to develop opponent modeling techniques for Poker agents. The approach is based on applying clustering algorithms to a Poker game database in order to identify player types based on their actions. First, common game moves were identified by clustering all players\' moves. Then, player types were defined by calculating the f…
Czech Technical University in Prague（布拉格捷克理工大学）；Nanyang Technological University（南洋理工大学）
Solution concepts of traditional game theory assume entirely rational players; therefore, their ability to exploit subrational opponents is limited. One type of subrationality that describes human behavior well is the quantal response. While there exist algorithms for computing solutions against quantal opponents, they either do not scale or may provide strategies that are even worse than the entirely-rational Nash strategies. This paper aims to analyze and propose scalable algorithms for computing effective and robust strategies against a quantal opponent in normal-form and extensive-form gam…
In the domain of the Soccer simulation 2D league of the RoboCup project, appropriate player positioning against a given opponent team is an important factor of soccer team performance. This work proposes a model which decides the strategy that should be applied regarding a particular opponent team. This task can be realized by applying preliminary a learning phase where the model determines the most effective strategies against clusters of opponent teams. The model determines the best strategies by using sequential Bayes' estimators. As a first trial of the system, the proposed model is used t…
The goal of this project is to predict the opponent's configuration in a RoboCup SSL environment. For simplicity, a Markov model assumption is made such that the predicted formation of the opponent team only depends on its current formation. The field is divided into a grid and a robot state per player is created with information about its position and its velocity. To gather a more general sense of what the opposing team is doing, the state also incorporates the team's average position (centroid). All possible state transitions are stored in a hash table that requires minimum storage space. T…
Poker is ideal for testing automated reasoning under uncertainty. It introduces uncertainty both by physical randomization and by incomplete information about opponents hands.Another source OF uncertainty IS the limited information available TO construct psychological models OF opponents, their tendencies TO bluff, play conservatively, reveal weakness, etc. AND the relation BETWEEN their hand strengths AND betting behaviour. ALL OF these uncertainties must be assessed accurately AND combined effectively FOR ANY reasonable LEVEL OF skill IN the game TO be achieved, since good decision making IS…
Basically, in (one-player) war Real Time Strategy (wRTS) games a human player controls, in real time, an army consisting of a number of soldiers and her aim is to destroy the opponent's assets where the opponent is a virtual (i.e., non-human player controlled) player that usually consists of a pre-programmed decision-making script. These scripts have usually associated some well-known problems (e.g., predictability, non-rationality, repetitive behaviors, and sensation of artificial stupidity among others). This paper describes a method for the automatic generation of virtual players that adapt…
Universidade Federal de Minas Gerais (UFMG)（米纳斯吉拉斯联邦大学）
This paper addresses the dynamic difficulty adjustment on MOBA games as a way to improve the player's entertainment. Although MOBA is currently one of the most played genres around the world, it is known as a game that offer less autonomy, more challenges and consequently more frustration. Due to these characteristics, the use of a mechanism that performs the difficulty balance dynamically seems to be an interesting alternative to minimize and/or avoid that players experience such frustrations. In this sense, this paper presents a dynamic difficulty adjustment mechanism for MOBA games. The mai…
Opponent Modelling tries to predict the future actions of opponents, and is required to perform well in multi-player games. There is a deep literature on learning an opponent model, but much less on how accurate such models must be to be useful. We investigate the sensitivity of Monte Carlo Tree Search (MCTS) and a Rolling Horizon Evolutionary Algorithm (RHEA) to the accuracy of their modelling of the opponent in a simple Real-Time Strategy game. We find that in this domain RHEA is much more sensitive to the accuracy of an opponent model than MCTS. MCTS generally does better even with an inacc…
An ideal strategy in zero-sum games should not only grant the player an average reward no less than the value of Nash equilibrium, but also exploit the (adaptive) opponents when they are suboptimal. While most existing works in Markov games focus exclusively on the former objective, it remains open whether we can achieve both objectives simultaneously. To address this problem, this work studies no-regret learning in Markov games with adversarial opponents when competing against the best fixed policy in hindsight. Along this direction, we present a new complete set of positive and negative resu…
Johns Hopkins University Applied Physics Laboratory（约翰霍普金斯大学应用物理实验室）
This paper provides a complexity analysis for the game of reconnaissance blind chess (RBC), a recently-introduced variant of chess where each player does not know the positions of the opponent's pieces a priori but may reveal a subset of them through chosen, private sensing actions. In contrast to many commonly studied imperfect information games like poker, an RBC player does not know what the opponent knows or has chosen to learn, exponentially expanding the size of the game's information sets (i.e., the number of possible game states that are consistent with what a player has observed). Eff…
Ganzfried Research（甘茨弗里德研究公司）；Florida International University（佛罗里达国际大学）
Creating strong agents for games with more than two players is a major open problem in AI. Common approaches are based on approximating game-theoretic solution concepts such as Nash equilibrium, which have strong theoretical guarantees in two-player zero-sum games, but no guarantees in non-zero-sum games or in games with more than two players. We describe an agent that is able to defeat a variety of realistic opponents using an exact Nash equilibrium strategy in a 3-player imperfect-information game. This shows that, despite a lack of theoretical guarantees, agents based on Nash equilibrium st…
Florida International University（佛罗里达国际大学）；Ganzfried Research（Ganzfried研究院）；Stanford University（斯坦福大学）
Two fundamental problems in computational game theory are computing a Nash equilibrium and learning to exploit opponents given observations of their play (opponent exploitation). The latter is perhaps even more important than the former: Nash equilibrium does not have a compelling theoretical justification in game classes other than two-player zero-sum, and for all games one can potentially do better by exploiting perceived weaknesses of the opponent than by following a static equilibrium strategy throughout the match. The natural setting for opponent exploitation is the Bayesian setting where…
We study infinite-horizon discounted two-player zero-sum Markov games, and develop a decentralized algorithm that provably converges to the set of Nash equilibria under self-play. Our algorithm is based on running an Optimistic Gradient Descent Ascent algorithm on each state to learn the policies, with a critic that slowly learns the value of each state. To the best of our knowledge, this is the first algorithm in this setting that is simultaneously rational (converging to the opponent's best response when it uses a stationary policy), convergent (converging to the set of Nash equilibria under…
Limited lookahead has been studied for decades in perfect-information games. We initiate a new direction via two simultaneous deviation points: generalization to imperfect-information games and a game-theoretic approach. We study how one should act when facing an opponent whose lookahead is limited. We study this for opponents that differ based on their lookahead depth, based on whether they, too, have imperfect information, and based on how they break ties. We characterize the hardness of finding a Nash equilibrium or an optimal commitment strategy for either player, showing that in some of t…
We introduce a class of extensive form games where players might not be able to foresee the possible consequences of their decisions and form a model of their opponents which they exploit to achieve a more profitable outcome. We improve upon existing models of games with limited foresight, endowing players with the ability of higher-order reasoning and proposing a novel solution concept to address intuitions coming from real game play. We analyse the resulting equilibria, devising an effective procedure to compute them.
We develop a new model that can be applied to any perfect information two-player zero-sum game to target a high score, and thus a perfect play. We integrate this model into the Monte Carlo tree search-policy iteration learning pipeline introduced by Google DeepMind with AlphaGo. Training this model on 9x9 Go produces a superhuman Go player, thus proving that it is stable and robust. We show that this model can be used to effectively play with both positional and score handicap, and to minimize suboptimal moves. We develop a family of agents that can target high scores against any opponent, and…
