arXiv subjects

Bora Yongacoglu

Publications and source records attributed to Bora Yongacoglu.

7 recordsLinked to original sources

arXiv2024

Paths to Equilibrium in Games

In multi-agent reinforcement learning (MARL) and game theory, agents repeatedly interact and revise their strategies as new data arrives, producing a sequence of strategy profiles. This paper studies sequences of strategies satisfying a pairwise constraint inspired by policy updating in reinforcement learning, where an agent who is best responding in one period does not switch its strategy in the next period. This constraint merely requires that optimizing agents do not switch strategies, but does not constrain the non-optimizing agents in any way, and thus allows for exploration. Sequences with this property are called satisficing paths, and arise naturally in many MARL algorithms. A fundamental question about strategic dynamics is such: for a given game and initial strategy profile, is it always possible to construct a satisficing path that terminates at an equilibrium? The resolution of this question has implications about the capabilities or limitations of a class of MARL algorithms. We answer this question in the affirmative for normal-form games. Our analysis reveals a counterintuitive insight that reward deteriorating strategic updates are key to driving play to equilibrium along a satisficing path.

cs.GT↗

arXiv2024

Satisficing Paths to Equilibrium, Generalized Weakly Acyclic Games, and Learning

Weakly acyclic games generalize potential games and have shown to be fundamental in the study of multi-agent learning as they allow for convergence to an equilibrium via best-responding under inertia. In this paper, we present a generalization of weakly acyclic games, and we demonstrate its importance in multi-agent learning when agents employ experimental strategy updates in periods where they fail to best respond. While weak acyclicity is defined in terms of path connectivity properties of a game's better response graph, our concept is defined using a generalized better response graph under revision dynamics termed as satisficing. We refer to this class of games as generalized weakly acyclic games (GenWAGs). We provide sufficient conditions for this notion of generalized weak acyclicity in both two-player games and n-player games in normal form, including static and dynamic games. Several graph theoretic characterizations of such games are presented together with sufficiency conditions, examples, and counterexamples. Finally, implications on learning via policy revision processes are presented.

cs.GT↗

arXiv2023

Unsynchronized Decentralized Q-Learning: Two Timescale Analysis By Persistence

Non-stationarity is a fundamental challenge in multi-agent reinforcement learning (MARL), where agents update their behaviour as they learn. Many theoretical advances in MARL avoid the challenge of non-stationarity by coordinating the policy updates of agents in various ways, including synchronizing times at which agents are allowed to revise their policies. Synchronization enables analysis of many MARL algorithms via multi-timescale methods, but such synchronization is infeasible in many decentralized applications. In this paper, we study an unsynchronized variant of the decentralized Q-learning algorithm, a recent MARL algorithm for stochastic games. We provide sufficient conditions under which the unsynchronized algorithm drives play to equilibrium with high probability. Our solution utilizes constant learning rates in the Q-factor update, which we show to be critical for relaxing the synchronization assumptions of earlier work. Our analysis also applies to unsynchronized generalizations of a number of other algorithms from the regret testing tradition, whose performance is analyzed by multi-timescale methods that study Markov chains obtained via policy update dynamics. This work extends the applicability of the decentralized Q-learning algorithm and its relatives to settings in which parameters are selected in an independent manner, and tames non-stationarity without imposing the coordination assumptions of prior work.

cs.GT↗

arXiv2023

Decentralized Multi-Agent Reinforcement Learning for Continuous-Space Stochastic Games

Stochastic games are a popular framework for studying multi-agent reinforcement learning (MARL). Recent advances in MARL have focused primarily on games with finitely many states. In this work, we study multi-agent learning in stochastic games with general state spaces and an information structure in which agents do not observe each other's actions. In this context, we propose a decentralized MARL algorithm and we prove the near-optimality of its policy updates. Furthermore, we study the global policy-updating dynamics for a general class of best-reply based algorithms and derive a closed-form characterization of convergence probabilities over the joint policy space.

cs.LG↗

arXiv2022

Mean-Field Games With Finitely Many Players: Independent Learning and Subjectivity

Independent learners are agents that employ single-agent algorithms in multi-agent systems, intentionally ignoring the effect of other strategic agents. This paper studies mean-field games from a decentralized learning perspective, with two primary objectives: (i) to identify structure that can guide algorithm design, and (ii) to understand the emergent behaviour in systems of independent learners. We study a new model of partially observed mean-field games with finitely many players, local action observability, and a general observation channel for partial observations of the global state. Specific observation channels considered include (a) global observability, (b) local and mean-field observability, (c) local and compressed mean-field observability, and (d) only local observability. We establish conditions under which the control problem of a given agent is equivalent to a fully observed MDP, as well as conditions under which the control problem is equivalent only to a POMDP. Building on the connection to MDPs, we prove the existence of perfect equilibrium among memoryless stationary policies under mean-field observability. Leveraging the connection to POMDPs, we prove convergence of learning iterates obtained by independent learning agents under any of the aforementioned observation channels. We interpret the limiting values as subjective value functions, which an agent believes to be relevant to its control problem. These subjective value functions are then used to propose subjective Q-equilibrium, a new solution concept for partially observed n-player mean-field games, whose existence is proved under mean-field or global observability. We provide a decentralized learning algorithm for partially observed n-player mean-field games, and we show that it drives play to subjective Q-equilibrium by adapting the recently developed theory of satisficing paths to allow for subjectivity.

cs.GT↗

arXiv2021

Satisficing Paths and Independent Multi-Agent Reinforcement Learning in Stochastic Games

In multi-agent reinforcement learning (MARL), independent learners are those that do not observe the actions of other agents in the system. Due to the decentralization of information, it is challenging to design independent learners that drive play to equilibrium. This paper investigates the feasibility of using satisficing dynamics to guide independent learners to approximate equilibrium in stochastic games. For $\epsilon \geq 0$, an $\epsilon$-satisficing policy update rule is any rule that instructs the agent to not change its policy when it is $\epsilon$-best-responding to the policies of the remaining players; $\epsilon$-satisficing paths are defined to be sequences of joint policies obtained when each agent uses some $\epsilon$-satisficing policy update rule to select its next policy. We establish structural results on the existence of $\epsilon$-satisficing paths into $\epsilon$-equilibrium in both symmetric $N$-player games and general stochastic games with two players. We then present an independent learning algorithm for $N$-player symmetric games and give high probability guarantees of convergence to $\epsilon$-equilibrium under self-play. This guarantee is made using symmetry alone, leveraging the previously unexploited structure of $\epsilon$-satisficing paths.

cs.GT↗

arXiv2019

Decentralized Learning for Optimality in Stochastic Dynamic Teams and Games with Local Control and Global State Information

Stochastic dynamic teams and games are rich models for decentralized systems and challenging testing grounds for multi-agent learning. Previous work that guaranteed team optimality assumed stateless dynamics, or an explicit coordination mechanism, or joint-control sharing. In this paper, we present an algorithm with guarantees of convergence to team optimal policies in teams and common interest games. The algorithm is a two-timescale method that uses a variant of Q-learning on the finer timescale to perform policy evaluation while exploring the policy space on the coarser timescale. Agents following this algorithm are "independent learners": they use only local controls, local cost realizations, and global state information, without access to controls of other agents. The results presented here are the first, to our knowledge, to give formal guarantees of convergence to team optimality using independent learners in stochastic dynamic teams and common interest games.

math.OC↗