A key challenge in applying reinforcement learning to safety-critical domains is understanding how to balance exploration (needed to attain good performance on the task) with safety (needed to avoid catastrophic failure). Although a growing line of work in reinforcement learning has investigated this area of "safe exploration," most existing techniques either 1) do not guarantee safety during the actual exploration process; and/or 2) limit the problem to a priori known and/or deterministic transition dynamics with strong smoothness assumptions. Addressing this gap, we propose Analogous Safe-state Exploration (ASE), an algorithm for provably safe exploration in MDPs with unknown, stochastic dynamics. Our method exploits analogies between state-action pairs to safely learn a near-optimal policy in a PAC-MDP sense. Additionally, ASE also guides exploration towards the most task-relevant states, which empirically results in significant improvements in terms of sample efficiency, when compared to existing methods.
Meta AI；UMPA, ENS Lyon（里昂高等师范学院纯数学与应用数学实验室）；Univ. Lille（里尔大学）；CNRS（法国国家科学研究中心）；Inria（法国国家信息与自动化研究所）；Centrale Lille（里尔中央理工学院）
In probably approximately correct (PAC) reinforcement learning (RL), an agent is required to identify an $ε$-optimal policy with probability $1-δ$. While minimax optimal algorithms exist for this problem, its instance-dependent complexity remains elusive in episodic Markov decision processes (MDPs). In this paper, we propose the first nearly matching (up to a horizon squared factor and logarithmic terms) upper and lower bounds on the sample complexity of PAC RL in deterministic episodic MDPs with finite state and action spaces. In particular, our bounds feature a new notion of sub-optimality g…
In this paper, we propose new problem-independent lower bounds on the sample complexity and regret in episodic MDPs, with a particular focus on the non-stationary case in which the transition kernel is allowed to change in each stage of the episode. Our main contribution is a novel lower bound of $Ω((H^3SA/ε^2)\log(1/δ))$ on the sample complexity of an $(\varepsilon,δ)$-PAC algorithm for best policy identification in a non-stationary MDP. This lower bound relies on a construction of "hard MDPs" which is different from the ones previously used in the literature. Using this same class of MDPs, w…
Australian National University（澳大利亚国立大学）；ETH Zürich（苏黎世联邦理工学院）；NICTA（国家信息与通信技术研究院）
We study upper and lower bounds on the sample-complexity of learning near-optimal behaviour in finite-state discounted Markov Decision Processes (MDPs). For the upper bound we make the assumption that each action leads to at most two possible next-states and prove a new bound for a UCRL-style algorithm on the number of time-steps when it is not Probably Approximately Correct (PAC). The new lower bound strengthens previous work by being both more general (it applies to all policies) and tighter. The upper and lower bounds match up to logarithmic factors.
In this paper we propose a method that learns to play Pac-Man. We define a set of high-level observation and action modules. Actions are temporally extended, and multiple action modules may be in effect concurrently. A decision of the agent is represented as a rule-based policy. For learning, we apply the cross-entropy method, a recent global optimization algorithm. The learned policies reached better score than the hand-crafted policy, and neared the score of average human players. We argue that learning is successful mainly because (i) the policy space includes the combination of individual …
Within the framework of probably approximately correct Markov decision processes (PAC-MDP), much theoretical work has focused on methods to attain near optimality after a relatively long period of learning and exploration. However, practical concerns require the attainment of satisfactory behavior within a short period of time. In this paper, we relax the PAC-MDP conditions to reconcile theoretically driven exploration methods and practical needs. We propose simple algorithms for discrete and continuous state spaces, and illustrate the benefits of our proposed relaxation via theoretical analys…
New York University（纽约大学）；Tata Institute of Fundamental Research（塔塔基础研究所）；Technical University of Munich（慕尼黑工业大学）；Chennai Mathematical Institute（金奈数学研究所）
Markov decision processes (MDP) and continuous-time MDP (CTMDP) are the fundamental models for non-deterministic systems with probabilistic uncertainty. Mean payoff (a.k.a. long-run average reward) is one of the most classic objectives considered in their context. We provide the first algorithm to compute mean payoff probably approximately correctly in unknown MDP; further, we extend it to unknown CTMDP. We do not require any knowledge of the state space, only a lower bound on the minimum transition probability, which has been advocated in literature. In addition to providing probably approxim…
A Policy Gradient Primal-Dual Algorithm for Constrained MDPs with Uniform PAC Guarantees
The University of Tokyo（东京大学）；OMRON SINIC X（欧姆龙SINIC X公司）；Mizuho–DL Financial Technology（瑞穗-DL金融科技公司）；Nara Institute of Science and Technology（奈良先端科学技术大学院大学）；RIKEN Center for Advanced Intelligence Project（理化学研究所革新智能综合研究中心）；Osaka University（大阪大学）；Kyoto University（京都大学）
Recently, there has been significant progress in understanding reinforcement learning in discounted infinite-horizon Markov decision processes (MDPs) by deriving tight sample complexity bounds. However, in many real-world applications, an interactive learning agent operates for a fixed or bounded period of time, for example tutoring students for exams or handling customer service requests. Such scenarios can often be better treated as episodic fixed-horizon MDPs, for which only looser bounds on the sample complexity exist. A natural notion of sample complexity in this setting is the number of …
Statistical performance bounds for reinforcement learning (RL) algorithms can be critical for high-stakes applications like healthcare. This paper introduces a new framework for theoretically measuring the performance of such algorithms called Uniform-PAC, which is a strengthening of the classical Probably Approximately Correct (PAC) framework. In contrast to the PAC framework, the uniform version may be used to derive high probability regret guarantees and so forms a bridge between the two setups that has been missing in the literature. We demonstrate the benefits of the new framework for fin…
Reward-free reinforcement learning (RL) considers the setting where the agent does not have access to a reward function during exploration, but must propose a near-optimal policy for an arbitrary reward function revealed only after exploring. In the the tabular setting, it is well known that this is a more difficult problem than reward-aware (PAC) RL -- where the agent has access to the reward function during exploration -- with optimal sample complexities in the two settings differing by a factor of $|\mathcal{S}|$, the size of the state space. We show that this separation does not exist in t…
ONERA, The French Aerospace Lab（法国航空航天研究院）；Brown University（布朗大学）；Amazon Web Service（亚马逊网络服务）
We consider the problem of knowledge transfer when an agent is facing a series of Reinforcement Learning (RL) tasks. We introduce a novel metric between Markov Decision Processes (MDPs) and establish that close MDPs have close optimal value functions. Formally, the optimal value functions are Lipschitz continuous with respect to the tasks space. These theoretical results lead us to a value-transfer method for Lifelong RL, which we use to build a PAC-MDP algorithm with improved convergence rate. Further, we show the method to experience no negative transfer with high probability. We illustrate …
MIT CSAIL（麻省理工学院计算机科学与人工智能实验室）；Brown University（布朗大学）
In recent years, researchers have made significant progress in devising reinforcement-learning algorithms for optimizing linear temporal logic (LTL) objectives and LTL-like objectives. Despite these advancements, there are fundamental limitations to how well this problem can be solved. Previous studies have alluded to this fact but have not examined it in depth. In this paper, we address the tractability of reinforcement learning for general LTL objectives from a theoretical perspective. We formalize the problem under the probably approximately correct learning in Markov decision processes (PA…
In the optimization of dynamic systems, the variables typically have constraints. Such problems can be modeled as a Constrained Markov Decision Process (CMDP). This paper considers the peak Constrained Markov Decision Process (PCMDP), where the agent chooses the policy to maximize total reward in the finite horizon as well as satisfy constraints at each epoch with probability 1. We propose a model-free algorithm that converts PCMDP problem to an unconstrained problem and a Q-learning based approach is applied. We define the concept of probably approximately correct (PAC) to the proposed PCMDP …
We study an exploration method for model-free RL that generalizes the counter-based exploration bonus methods and takes into account long term exploratory value of actions rather than a single step look-ahead. We propose a model-free RL method that modifies Delayed Q-learning and utilizes the long-term exploration bonus with provable efficiency. We show that our proposed method finds a near-optimal policy in polynomial time (PAC-MDP), and also provide experimental evidence that our proposed algorithm is an efficient exploration method.
University of Southern California（南加州大学）；Indian Institute of Technology, Kharagpur（印度理工学院卡拉格普尔分校）；Adobe MDSR
Humans excel at learning long-horizon tasks from demonstrations augmented with textual commentary, as evidenced by the burgeoning popularity of tutorial videos online. Intuitively, this capability can be separated into 2 distinct subtasks - first, dividing a long-horizon demonstration sequence into semantically meaningful events; second, adapting such events into meaningful behaviors in one's own environment. Here, we present Video2Skill (V2S), which attempts to extend this capability to artificial agents by allowing a robot arm to learn from human cooking videos. We first use sequence-to-sequ…
Carnegie Mellon University（卡内基梅隆大学）；University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；Microsoft Research（微软研究院）
We study the sample complexity of model-based reinforcement learning (henceforth RL) in general contextual decision processes that require strategic exploration to find a near-optimal policy. We design new algorithms for RL with a generic model class and analyze their statistical properties. Our algorithms have sample complexity governed by a new structural parameter called the witness rank, which we show to be small in several settings of interest, including factored MDPs. We also show that the witness rank is never larger than the recently proposed Bellman rank parameter governing the sample…
Missingness-MDPs: Bridging the Theory of Missing Data and POMDPs
Ruhr University Bochum（博德姆鲁尔大学）；Brno University of Technology（布拉格技术大学）；Radboud University Nijmegen（拉德博德大学奈杰姆）；Harvard University（哈佛大学）；Eindhoven University of Technology（埃因霍温理工大学）
This paper offers a new hybrid probably approximately correct (PAC) reinforcement learning (RL) algorithm for Markov decision processes (MDPs) that intelligently maintains favorable features of its parents. The designed algorithm, referred to as the Dyna-Delayed Q-learning (DDQ) algorithm, combines model-free and model-based learning approaches while outperforming both in most cases. The paper includes a PAC analysis of the DDQ algorithm and a derivation of its sample complexity. Numerical results are provided to support the claim regarding the new algorithm's sample efficiency compared to its…
We propose a framework for transferring any existing policy from a potentially unknown source MDP to a target MDP. This framework (1) enables reuse in the target domain of any form of source policy, including classical controllers, heuristic policies, or deep neural network-based policies, (2) attains optimality under suitable theoretical conditions, and (3) guarantees improvement over the source policy in the target MDP. These are achieved by packaging the source policy as a black-box option in the target MDP and providing a theoretically grounded way to learn the option's initiation set thro…
We develop model free PAC performance guarantees for multiple concurrent MDPs, extending recent works where a single learner interacts with multiple non-interacting agents in a noise free environment. Our framework allows noisy and resource limited communication between agents, and develops novel PAC guarantees in this extended setting. By allowing communication between the agents themselves, we suggest improved PAC-exploration algorithms that can overcome the communication noise and lead to improved sample complexity bounds. We provide a theoretically motivated algorithm that optimally combin…
Efficient exploration is one of the key challenges for reinforcement learning (RL) algorithms. Most traditional sample efficiency bounds require strategic exploration. Recently many deep RL algorithms with simple heuristic exploration strategies that have few formal guarantees, achieve surprising success in many domains. These results pose an important question about understanding these exploration strategies such as $e$-greedy, as well as understanding what characterize the difficulty of exploration in MDPs. In this work we propose problem specific sample complexity bounds of $Q$ learning wit…
Many physical systems have underlying safety considerations that require that the policy employed ensures the satisfaction of a set of constraints. The analytical formulation usually takes the form of a Constrained Markov Decision Process (CMDP). We focus on the case where the CMDP is unknown, and RL algorithms obtain samples to discover the model and compute an optimal constrained policy. Our goal is to characterize the relationship between safety constraints and the number of samples needed to ensure a desired level of accuracy -- both objective maximization and constraint satisfaction -- in…
Meta AI；ENS Lyon（里昂高等师范学院）；Univ. Lille（里尔大学）；Inria（法国国家信息与自动化研究所）；CNRS（法国国家科学研究中心）；Centrale Lille（里尔中央理工学院）
Optimistic algorithms have been extensively studied for regret minimization in episodic tabular MDPs, both from a minimax and an instance-dependent view. However, for the PAC RL problem, where the goal is to identify a near-optimal policy with high probability, little is known about their instance-dependent sample complexity. A negative result of Wagenmaker et al. (2021) suggests that optimistic sampling rules cannot be used to attain the (still elusive) optimal instance-dependent sample complexity. On the positive side, we provide the first instance-dependent bound for an optimistic algorithm…
Technion – Israel Institute of Technology（以色列理工学院）
In the Bayesian reinforcement learning (RL) setting, a prior distribution over the unknown problem parameters -- the rewards and transitions -- is assumed, and a policy that optimizes the (posterior) expected return is sought. A common approximation, which has been recently popularized as meta-RL, is to train the agent on a sample of $N$ problem instances from the prior, with the hope that for large enough $N$, good generalization behavior to an unseen test instance will be obtained. In this work, we study generalization in Bayesian RL under the probably approximately correct (PAC) framework, …
Constrained Markov Decision Processes (CMDPs) formalize sequential decision-making problems whose objective is to minimize a cost function while satisfying constraints on various cost functions. In this paper, we consider the setting of episodic fixed-horizon CMDPs. We propose an online algorithm which leverages the linear programming formulation of finite-horizon CMDP for repeated optimistic planning to provide a probably approximately correct (PAC) guarantee on the number of episodes needed to ensure an $ε$-optimal policy, i.e., with resulting objective value within $ε$ of the optimal value …
VTT Technical Research Centre of Finland（芬兰VTT技术研究中心）；University of Electro-Communications（电气通信大学）；Zhejiang University（浙江大学）；University of Oslo（奥斯陆大学）；University of Oulu（奥卢大学）
This paper targets at the problem of radio resource management for expected long-term delay-power tradeoff in vehicular communications. At each decision epoch, the road side unit observes the global network state, allocates channels and schedules data packets for all vehicle user equipment-pairs (VUE-pairs). The decision-making procedure is modelled as a discrete-time Markov decision process (MDP). The technical challenges in solving an optimal control policy originate from highly spatial mobility of vehicles and temporal variations in data traffic. To simplify the decision-making process, we …
Stanford University（斯坦福大学）；University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；UC Santa Barbara（加州大学圣巴巴拉分校）
We take initial steps in studying PAC-MDP algorithms with limited adaptivity, that is, algorithms that change its exploration policy as infrequently as possible during regret minimization. This is motivated by the difficulty of running fully adaptive algorithms in real-world applications (such as medical domains), and we propose to quantify adaptivity using the notion of local switching cost. Our main contribution, Q-Learning with UCB2 exploration, is a model-free algorithm for H-step episodic MDP that achieves sublinear regret whose local switching cost in K episodes is $O(H^3SA\log K)$, and …
VTT Technical Research Centre of Finland Ltd（芬兰VTT技术研究中心）；University of Electro-Communications（电气通信大学）；Zhejiang University（浙江大学）；Shizuoka University（静冈大学）；University of Oslo（奥斯陆大学）；University of Oulu（奥卢大学）
In this paper, we investigate the problem of age of information (AoI)-aware radio resource management for expected long-term performance optimization in a Manhattan grid vehicle-to-vehicle network. With the observation of global network state at each scheduling slot, the roadside unit (RSU) allocates the frequency bands and schedules packet transmissions for all vehicle user equipment-pairs (VUE-pairs). We model the stochastic decision-making procedure as a discrete-time single-agent Markov decision process (MDP). The technical challenges in solving the optimal control policy originate from hi…
Carnegie Mellon University（卡内基梅隆大学）；University of California, Los Angles（加利福尼亚大学洛杉矶分校）
Recently there is a surge of interest in understanding the horizon-dependence of the sample complexity in reinforcement learning (RL). Notably, for an RL environment with horizon length $H$, previous work have shown that there is a probably approximately correct (PAC) algorithm that learns an $O(1)$-optimal policy using $\mathrm{polylog}(H)$ episodes of environment interactions when the number of states and actions is fixed. It is yet unknown whether the $\mathrm{polylog}(H)$ dependence is necessary or not. In this work, we resolve this question by developing an algorithm that achieves the sam…
Department of Computer Science, University of Manchester（曼彻斯特大学计算机科学系）；Faculty of Computer Science & DSME, RWTH Aachen University（亚琛工业大学计算机科学与DSME学院）；Department of Computer Science, University of Oxford（牛津大学计算机科学系）；Department of Computing, Imperial College London（伦敦帝国理工学院计算系）
The theory of reinforcement learning has focused on two fundamental problems: achieving low regret, and identifying $ε$-optimal policies. While a simple reduction allows one to apply a low-regret algorithm to obtain an $ε$-optimal policy and achieve the worst-case optimal rate, it is unknown whether low-regret algorithms can obtain the instance-optimal rate for policy identification. We show this is not possible -- there exists a fundamental tradeoff between achieving low regret and identifying an $ε$-optimal policy at the instance-optimal rate. Motivated by our negative finding, we propose a …
What Are the Odds? Improving the foundations of Statistical Model Checking
Lancaster University Leipzig（莱斯特大学莱比锡分校）；Ruhr University Bochum（波鸿鲁尔大学）；Institute of Science and Technology Austria（奥地利科学技术研究所）；Dresden University of Technology（德累斯顿工业大学）；Centre for Tactile Internet with Human-in-the-Loop (CeTI)（人机交互触觉互联网中心（CeTI））
Federated Learning With Energy Harvesting Devices: An MDP Framework
Department of Electronic and Computer Engineering, The Hong Kong University of Science and Technology（香港科技大学电子与计算机工程系）；School of Electrical Engineering and Computer Science, Washington State University（华盛顿州立大学电气与计算机科学学院）
This paper targets the efficient construction of a safety shield for decision making in scenarios that incorporate uncertainty. Markov decision processes (MDPs) are prominent models to capture such planning problems. Reinforcement learning (RL) is a machine learning technique to determine near-optimal policies in MDPs that may be unknown prior to exploring the model. However, during exploration, RL is prone to induce behavior that is undesirable or not allowed in safety- or mission-critical contexts. We introduce the concept of a probabilistic shield that enables decision-making to adhere to s…
We study model-based offline Reinforcement Learning with general function approximation without a full coverage assumption on the offline data distribution. We present an algorithm named Constrained Pessimistic Policy Optimization (CPPO)which leverages a general function class and uses a constraint over the model class to encode pessimism. Under the assumption that the ground truth model belongs to our function class (i.e., realizability in the function class), CPPO has a PAC guarantee with offline data only providing partial coverage, i.e., it can learn a policy that competes against any poli…
Qatar Computing Research Institute（卡塔尔计算研究所）；Hamad Bin Khalifa University（哈马德·本·哈利法大学）
Traffic signal control is an important problem in urban mobility with a significant potential of economic and environmental impact. While there is a growing interest in Reinforcement Learning (RL) for traffic signal control, the work so far has focussed on learning through simulations which could lead to inaccuracies due to simplifying assumptions. Instead, real experience data on traffic is available and could be exploited at minimal costs. Recent progress in offline or batch RL has enabled just that. Model-based offline RL methods, in particular, have been shown to generalize from the experi…
Radboud University Nijmegen（拉德堡德大学奈梅亨分校）；INRIA Lille（法国国家信息与自动化研究所里尔分部）
We consider the problem of learning the optimal action-value function in the discounted-reward Markov decision processes (MDPs). We prove a new PAC bound on the sample-complexity of model-based value iteration algorithm in the presence of the generative model, which indicates that for an MDP with N state-action pairs and the discount factor γ\in[0,1) only O(N\log(N/δ)/((1-γ)^3ε^2)) samples are required to find an ε-optimal estimation of the action-value function with the probability 1-δ. We also prove a matching lower bound of Θ(N\log(N/δ)/((1-γ)^3ε^2)) on the sample complexity of estimating t…
The Ohio State University（俄亥俄州立大学）；Princeton University（普林斯顿大学）；Northwestern University（西北大学）
Designing off-policy reinforcement learning algorithms is typically a very challenging task, because a desirable iteration update often involves an expectation over an on-policy distribution. Prior off-policy actor-critic (AC) algorithms have introduced a new critic that uses the density ratio for adjusting the distribution mismatch in order to stabilize the convergence, but at the cost of potentially introducing high biases due to the estimation errors of both the density ratio and value function. In this paper, we develop a doubly robust off-policy AC (DR-Off-PAC) for discounted MDP, which c…
There has been growing progress on theoretical analyses for provably efficient learning in MDPs with linear function approximation, but much of the existing work has made strong assumptions to enable exploration by conventional exploration frameworks. Typically these assumptions are stronger than what is needed to find good solutions in the batch setting. In this work, we show how under a more standard notion of low inherent Bellman error, typically employed in least-square value iteration-style algorithms, we can provide strong PAC guarantees on learning a near optimal value function provided…
University of Technology Sydney（悉尼科技大学）；City University of Macau（澳门城市大学）
Reinforcement learning (RL) is one of the most important branches of AI. Due to its capacity for self-adaption and decision-making in dynamic environments, reinforcement learning has been widely applied in multiple areas, such as healthcare, data markets, autonomous driving, and robotics. However, some of these applications and systems have been shown to be vulnerable to security or privacy attacks, resulting in unreliable or unstable services. A large number of studies have focused on these security and privacy problems in reinforcement learning. However, few surveys have provided a systemati…
The performance of a reinforcement learning algorithm can vary drastically during learning because of exploration. Existing algorithms provide little information about the quality of their current policy before executing it, and thus have limited use in high-stakes applications like healthcare. We address this lack of accountability by proposing that algorithms output policy certificates. These certificates bound the sub-optimality and return of the policy in the next episode, allowing humans to intervene when the certified quality is not satisfactory. We further introduce two new algorithms w…
Achieving efficient and scalable exploration in complex domains poses a major challenge in reinforcement learning. While Bayesian and PAC-MDP approaches to the exploration problem offer strong formal guarantees, they are often impractical in higher dimensions due to their reliance on enumerating the state-action space. Hence, exploration in complex domains is often performed with simple epsilon-greedy methods. In this paper, we consider the challenging Atari games domain, which requires processing raw pixel inputs and delayed rewards. We evaluate several more sophisticated exploration strategi…
Ambient backscatter has been introduced with a wide range of applications for low power wireless communications. In this article, we propose an optimal and low-complexity dynamic spectrum access framework for RF-powered ambient backscatter system. In this system, the secondary transmitter not only harvests energy from ambient signals (from incumbent users), but also backscatters these signals to its receiver for data transmission. Under the dynamics of the ambient signals, we first adopt the Markov decision process (MDP) framework to obtain the optimal policy for the secondary transmitter, aim…
University of Pisa（比萨大学）；University of Chieti-Pescara（基耶蒂-佩斯卡拉大学）
Having access to an exploring restart distribution (the so-called wide coverage assumption) is critical with policy gradient methods. This is due to the fact that, while the objective function is insensitive to updates in unlikely states, the agent may still need improvements in those states in order to reach a nearly optimal payoff. For this reason, wide coverage is used in some form when analyzing theoretical properties of practical policy gradient methods. However, this assumption can be unfeasible in certain environments, for instance when learning is online, or when restarts are possible …
