Actor-Critics Can Achieve Optimal Sample Efficiency

• URL: http://arxiv.org/abs/2505.03710v1
• Date: Tue, 06 May 2025 17:32:39 GMT
• Title: Actor-Critics Can Achieve Optimal Sample Efficiency
• Authors: Kevin Tan, Wei Fan, Yuting Wei,
• Abstract summary: We introduce a novel actor-critic algorithm that attains a sample-complexity of $O(dH5 log|mathcalA|/epsilon2 + d H4 log|mathcalF|/ epsilon2)$ trajectories.<n>We extend this to the setting of Hybrid RL, showing that initializing the critic with offline data yields sample efficiency gains compared to purely offline or online RL.
• Score: 15.033410073144939
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Actor-critic algorithms have become a cornerstone in reinforcement learning (RL), leveraging the strengths of both policy-based and value-based methods. Despite recent progress in understanding their statistical efficiency, no existing work has successfully learned an $\epsilon$-optimal policy with a sample complexity of $O(1/\epsilon^2)$ trajectories with general function approximation when strategic exploration is necessary. We address this open problem by introducing a novel actor-critic algorithm that attains a sample-complexity of $O(dH^5 \log|\mathcal{A}|/\epsilon^2 + d H^4 \log|\mathcal{F}|/ \epsilon^2)$ trajectories, and accompanying $\sqrt{T}$ regret when the Bellman eluder dimension $d$ does not increase with $T$ at more than a $\log T$ rate. Here, $\mathcal{F}$ is the critic function class, $\mathcal{A}$ is the action space, and $H$ is the horizon in the finite horizon MDP setting. Our algorithm integrates optimism, off-policy critic estimation targeting the optimal Q-function, and rare-switching policy resets. We extend this to the setting of Hybrid RL, showing that initializing the critic with offline data yields sample efficiency gains compared to purely offline or online RL. Further, utilizing access to offline data, we provide a \textit{non-optimistic} provably efficient actor-critic algorithm that only additionally requires $N_{\text{off}} \geq c_{\text{off}}^*dH^4/\epsilon^2$ in exchange for omitting optimism, where $c_{\text{off}}^*$ is the single-policy concentrability coefficient and $N_{\text{off}}$ is the number of offline samples. This addresses another open problem in the literature. We further provide numerical experiments to support our theoretical findings.

Related papers

• Multiple-policy Evaluation via Density Estimation [30.914344538340412]
  We propose an algorithm named $mathrmCAESAR$ for this problem. Up to low order and logarithmic terms $mathrmCAESAR$ achieves a sample complexity $tildeOleft(fracH4epsilon2sum_h=1Hmax_kin[K]sum_s,afrac(d_hpik(s,a))2mu*_h(s,a)right)$, where $dpi
  arXiv  Detail & Related papers  (2024-03-29T23:55:25Z)
• Sharper Model-free Reinforcement Learning for Average-reward Markov Decision Processes [21.77276136591518]
  We develop provably efficient model-free reinforcement learning (RL) algorithms for Markov Decision Processes (MDPs) In the simulator setting, we propose a model-free RL algorithm that finds an $epsilon$-optimal policy using $widetildeO left(fracSAmathrmsp(h*)epsilon2+fracS2Amathrmsp(h*)epsilon2right)$ samples.
  arXiv  Detail & Related papers  (2023-06-28T17:43:19Z)
• Eluder-based Regret for Stochastic Contextual MDPs [43.19667415823089]
  We present the E-UC$3$RL algorithm for regret minimization in Contextual Markov Decision Processes (CMDPs) Our algorithm is efficient (assuming efficient offline regression oracles) and enjoys a regret guarantee of $ widetildeO(H3 sqrtT |S| |A|d_mathrmE(mathcalP)$.
  arXiv  Detail & Related papers  (2022-11-27T20:38:47Z)
• Provably Efficient Offline Reinforcement Learning with Trajectory-Wise Reward [66.81579829897392]
  We propose a novel offline reinforcement learning algorithm called Pessimistic vAlue iteRaTion with rEward Decomposition (PARTED) PARTED decomposes the trajectory return into per-step proxy rewards via least-squares-based reward redistribution, and then performs pessimistic value based on the learned proxy reward. To the best of our knowledge, PARTED is the first offline RL algorithm that is provably efficient in general MDP with trajectory-wise reward.
  arXiv  Detail & Related papers  (2022-06-13T19:11:22Z)
• Nearly Optimal Policy Optimization with Stable at Any Time Guarantee [53.155554415415445]
  Policy-based method in citetshani 2020optimistic is only $tildeO(sqrtSAH3K + sqrtAH4K)$ where $S$ is the number of states, $A$ is the number of actions, $H$ is the horizon, and $K$ is the number of episodes, and there is a $sqrtSH$ gap compared with the information theoretic lower bound $tildeOmega(sqrtSAH
  arXiv  Detail & Related papers  (2021-12-21T01:54:17Z)
• Towards Instance-Optimal Offline Reinforcement Learning with Pessimism [34.54294677335518]
  We study the offline reinforcement learning (offline RL) problem, where the goal is to learn a reward-maximizing policy in an unknown Markov Decision Process (MDP) In this work, we analyze the Adaptive Pessimistic Value Iteration (APVI) algorithm and derive the suboptimality upper bound that nearly matches [ Oleft(sum_h=1Hsum_s_h,a_hdpistar_h(s_h,a_h)sqrtfracmathrmmathrmVar_
  arXiv  Detail & Related papers  (2021-10-17T01:21:52Z)
• Gap-Dependent Unsupervised Exploration for Reinforcement Learning [40.990467706237396]
  We present an efficient algorithm for task-agnostic reinforcement learning. The algorithm takes only $widetildemathcalO (1/epsilon cdot (H3SA / rho + H4 S2 A) )$ episodes of exploration. We show that, information-theoretically, this bound is nearly tight for $rho Theta (1/(HS))$ and $H>1$.
  arXiv  Detail & Related papers  (2021-08-11T20:42:46Z)
• The Curse of Passive Data Collection in Batch Reinforcement Learning [82.6026077420886]
  In high stake applications, active experimentation may be considered too risky and thus data are often collected passively. While in simple cases, such as in bandits, passive and active data collection are similarly effective, the price of passive sampling can be much higher when collecting data from a system with controlled states.
  arXiv  Detail & Related papers  (2021-06-18T07:54:23Z)
• Randomized Exploration for Reinforcement Learning with General Value Function Approximation [122.70803181751135]
  We propose a model-free reinforcement learning algorithm inspired by the popular randomized least squares value iteration (RLSVI) algorithm. Our algorithm drives exploration by simply perturbing the training data with judiciously chosen i.i.d. scalar noises. We complement the theory with an empirical evaluation across known difficult exploration tasks.
  arXiv  Detail & Related papers  (2021-06-15T02:23:07Z)
• Online Sub-Sampling for Reinforcement Learning with General Function Approximation [111.01990889581243]
  In this paper, we establish an efficient online sub-sampling framework that measures the information gain of data points collected by an RL algorithm. For a value-based method with complexity-bounded function class, we show that the policy only needs to be updated for $proptooperatornamepolylog(K)$ times. In contrast to existing approaches that update the policy for at least $Omega(K)$ times, our approach drastically reduces the number of optimization calls in solving for a policy.
  arXiv  Detail & Related papers  (2021-06-14T07:36:25Z)
• Policy Finetuning: Bridging Sample-Efficient Offline and Online Reinforcement Learning [59.02541753781001]
  This paper initiates the theoretical study of policy finetuning, that is, online RL where the learner has additional access to a "reference policy" We first design a sharp offline reduction algorithm that finds an $varepsilon$ near-optimal policy within $widetildeO(H3SCstar/varepsilon2)$ episodes. We then establish an $Omega(H3SminCstar, A/varepsilon2)$ sample complexity lower bound for any policy finetuning algorithm, including those that can adaptively explore the
  arXiv  Detail & Related papers  (2021-06-09T08:28:55Z)
• Optimal Regret Algorithm for Pseudo-1d Bandit Convex Optimization [51.23789922123412]
  We study online learning with bandit feedback (i.e. learner has access to only zeroth-order oracle) where cost/reward functions admit a "pseudo-1d" structure. We show a lower bound of $min(sqrtdT, T3/4)$ for the regret of any algorithm, where $T$ is the number of rounds. We propose a new algorithm sbcalg that combines randomized online gradient descent with a kernelized exponential weights method to exploit the pseudo-1d structure effectively.
  arXiv  Detail & Related papers  (2021-02-15T08:16:51Z)
• Nearly Minimax Optimal Reinforcement Learning for Linear Mixture Markov Decision Processes [91.38793800392108]
  We study reinforcement learning with linear function approximation where the underlying transition probability kernel of the Markov decision process (MDP) is a linear mixture model. We propose a new, computationally efficient algorithm with linear function approximation named $textUCRL-VTR+$ for the aforementioned linear mixture MDPs. To the best of our knowledge, these are the first computationally efficient, nearly minimax optimal algorithms for RL with linear function approximation.
  arXiv  Detail & Related papers  (2020-12-15T18:56:46Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.