Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userConcave Utility Reinforcement Learning: the Mean-field Game viewpoint
252
0
0.0
(
0
)
Added by
Matthieu Geist
Publication date
2021
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
Concave Utility Reinforcement Learning (CURL) extends RL from linear to concave utilities in the occupancy measure induced by the agents policy. This encompasses not only RL but also imitation learning and exploration, among others. Yet, this more general paradigm invalidates the classical Bellman equations, and calls for new algorithms. Mean-field Games (MFGs) are a continuous approximation of many-agent RL. They consider the limit case of a continuous distribution of identical agents, anonymous with symmetric interests, and reduce the problem to the study of a single representative agent in interaction with the full population. Our core contribution consists in showing that CURL is a subclass of MFGs. We think this important to bridge together both communities. It also allows to shed light on aspects of both fields: we show the equivalence between concavity in CURL and monotonicity in the associated MFG, between optimality conditions in CURL and Nash equilibrium in MFG, or that Fictitious Play (FP) for this class of MFGs is simply Frank-Wolfe, bringing the first convergence rate for discrete-time FP for MFGs. We also experimentally demonstrate that, using algorithms recently introduced for solving MFGs, we can address the CURL problem more efficiently.
rate research
Read More
We consider the problem of tabular infinite horizon concave utility reinforcement learning (CURL) with convex constraints. Various learning applications with constraints, such as robotics, do not allow for policies that can violate constraints. To this end, we propose a model-based learning algorithm that achieves zero constraint violations. To obtain this result, we assume that the concave objective and the convex constraints have a solution interior to the set of feasible occupation measures. We then solve a tighter optimization problem to ensure that the constraints are never violated despite the imprecise model knowledge and model stochasticity. We also propose a novel Bellman error based analysis for tabular infinite-horizon setups which allows to analyse stochastic policies. Combining the Bellman error based analysis and tighter optimization equation, for $T$ interactions with the environment, we obtain a regret guarantee for objective which grows as $Tilde{O}(1/sqrt{T})$, excluding other factors.
We present a Reinforcement Learning (RL) algorithm to solve infinite horizon asymptotic Mean Field Game (MFG) and Mean Field Control (MFC) problems. Our approach can be described as a unified two-timescale Mean Field Q-learning: The emph{same} algorithm can learn either the MFG or the MFC solution by simply tuning the ratio of two learning parameters. The algorithm is in discrete time and space where the agent not only provides an action to the environment but also a distribution of the state in order to take into account the mean field feature of the problem. Importantly, we assume that the agent can not observe the populations distribution and needs to estimate it in a model-free manner. The asymptotic MFG and MFC problems are also presented in continuous time and space, and compared with classical (non-asymptotic or stationary) MFG and MFC problems. They lead to explicit solutions in the linear-quadratic (LQ) case that are used as benchmarks for the results of our algorithm.
Agents that interact with other agents often do not know a priori what the other agents strategies are, but have to maximise their own online return while interacting with and learning about others. The optimal adaptive behaviour under uncertainty over the other agents strategies w.r.t. some prior can in principle be computed using the Interactive Bayesian Reinforcement Learning framework. Unfortunately, doing so is intractable in most settings, and existing approximation methods are restricted to small tasks. To overcome this, we propose to meta-learn approximate belief inference and Bayes-optimal behaviour for a given prior. To model beliefs over other agents, we combine sequential and hierarchical Variational Auto-Encoders, and meta-train this inference model alongside the policy. We show empirically that our approach outperforms existing methods that use a model-free approach, sample from the approximate posterior, maintain memory-free models of others, or do not fully utilise the known structure of the environment.
One of the challenges for multi-agent reinforcement learning (MARL) is designing efficient learning algorithms for a large system in which each agent has only limited or partial information of the entire system. In this system, it is desirable to learn policies of a decentralized type. A recent and promising paradigm to analyze such decentralized MARL is to take network structures into consideration. While exciting progress has been made to analyze decentralized MARL with the network of agents, often found in social networks and team video games, little is known theoretically for decentralized MARL with the network of states, frequently used for modeling self-driving vehicles, ride-sharing, and data and traffic routing. This paper proposes a framework called localized training and decentralized execution to study MARL with network of states, with homogeneous (a.k.a. mean-field type) agents. Localized training means that agents only need to collect local information in their neighboring states during the training phase; decentralized execution implies that, after the training stage, agents can execute the learned decentralized policies, which only requires knowledge of the agents current states. The key idea is to utilize the homogeneity of agents and regroup them according to their states, thus the formulation of a networked Markov decision process with teams of agents, enabling the update of the Q-function in a localized fashion. In order to design an efficient and scalable reinforcement learning algorithm under such a framework, we adopt the actor-critic approach with over-parameterized neural networks, and establish the convergence and sample complexity for our algorithm, shown to be scalable with respect to the size of both agents and states.
Reinforcement Learning (RL) is an effective tool for controller design but can struggle with issues of robustness, failing catastrophically when the underlying system dynamics are perturbed. The Robust RL formulation tackles this by adding worst-case adversarial noise to the dynamics and constructing the noise distribution as the solution to a zero-sum minimax game. However, existing work on learning solutions to the Robust RL formulation has primarily focused on training a single RL agent against a single adversary. In this work, we demonstrate that using a single adversary does not consistently yield robustness to dynamics variations under standard parametrizations of the adversary; the resulting policy is highly exploitable by new adversaries. We propose a population-based augmentation to the Robust RL formulation in which we randomly initialize a population of adversaries and sample from the population uniformly during training. We empirically validate across robotics benchmarks that the use of an adversarial population results in a more robust policy that also improves out-of-distribution generalization. Finally, we demonstrate that this approach provides comparable robustness and generalization as domain randomization on these benchmarks while avoiding a ubiquitous domain randomization failure mode.
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
