Randomized Exploration for Reinforcement Learning with General Value Function Approximation

• URL: http://arxiv.org/abs/2106.07841v1
• Date: Tue, 15 Jun 2021 02:23:07 GMT
• Title: Randomized Exploration for Reinforcement Learning with General Value Function Approximation
• Authors: Haque Ishfaq, Qiwen Cui, Viet Nguyen, Alex Ayoub, Zhuoran Yang, Zhaoran Wang, Doina Precup, Lin F. Yang
• Abstract summary: 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.
• Score: 122.70803181751135
• License: http://creativecommons.org/publicdomain/zero/1.0/
• Abstract: We propose a model-free reinforcement learning algorithm inspired by the popular randomized least squares value iteration (RLSVI) algorithm as well as the optimism principle. Unlike existing upper-confidence-bound (UCB) based approaches, which are often computationally intractable, our algorithm drives exploration by simply perturbing the training data with judiciously chosen i.i.d. scalar noises. To attain optimistic value function estimation without resorting to a UCB-style bonus, we introduce an optimistic reward sampling procedure. When the value functions can be represented by a function class $\mathcal{F}$, our algorithm achieves a worst-case regret bound of $\widetilde{O}(\mathrm{poly}(d_EH)\sqrt{T})$ where $T$ is the time elapsed, $H$ is the planning horizon and $d_E$ is the $\textit{eluder dimension}$ of $\mathcal{F}$. In the linear setting, our algorithm reduces to LSVI-PHE, a variant of RLSVI, that enjoys an $\widetilde{\mathcal{O}}(\sqrt{d^3H^3T})$ regret. We complement the theory with an empirical evaluation across known difficult exploration tasks.

Related papers

• Provably Efficient Infinite-Horizon Average-Reward Reinforcement Learning with Linear Function Approximation [1.8416014644193066]
  We propose an algorithm for learning linear Markov decision processes (MDPs) and linear mixture MDPs under the Bellman optimality condition. Our algorithm for linear MDPs achieves the best-known regret upper bound of $widetildemathcalO(d3/2mathrmsp(v*)sqrtT)$ over $T$ time steps. For linear mixture MDPs, our algorithm attains a regret bound of $widetildemathcalO(dcdotmathrm
  arXiv  Detail & Related papers  (2024-09-16T23:13:42Z)
• Achieving Tractable Minimax Optimal Regret in Average Reward MDPs [19.663336027878408]
  We present the first tractable algorithm with minimax optimal regret of $widetildemathrmO(sqrtmathrmsp(h*) S A T)$. Remarkably, our algorithm does not require prior information on $mathrmsp(h*)$.
  arXiv  Detail & Related papers  (2024-06-03T11:53:44Z)
• Nearly Minimax Optimal Regret for Learning Linear Mixture Stochastic Shortest Path [80.60592344361073]
  We study the Shortest Path (SSP) problem with a linear mixture transition kernel. An agent repeatedly interacts with a environment and seeks to reach certain goal state while minimizing the cumulative cost. Existing works often assume a strictly positive lower bound of the iteration cost function or an upper bound of the expected length for the optimal policy.
  arXiv  Detail & Related papers  (2024-02-14T07:52:00Z)
• Nearly Minimax Optimal Reinforcement Learning with Linear Function Approximation [25.60689712525918]
  We study reinforcement learning with linear function approximation where the transition probability and reward functions are linear. We propose a novel-efficient algorithm, LSVI-UCB$+$, which achieves an $widetildeO(HdsqrtT)$ regret bound where $H$ is the episode length, $d$ is the feature dimension, and $T$ is the number of steps.
  arXiv  Detail & Related papers  (2022-06-23T06:04:21Z)
• Bayesian Optimistic Optimisation with Exponentially Decaying Regret [58.02542541410322]
  The current practical BO algorithms have regret bounds ranging from $mathcalO(fraclogNsqrtN)$ to $mathcal O(e-sqrtN)$, where $N$ is the number of evaluations. This paper explores the possibility of improving the regret bound in the noiseless setting by intertwining concepts from BO and tree-based optimistic optimisation. We propose the BOO algorithm, a first practical approach which can achieve an exponential regret bound with order $mathcal O(N-sqrt
  arXiv  Detail & Related papers  (2021-05-10T13:07:44Z)
• Randomized Exploration is Near-Optimal for Tabular MDP [45.16374124699648]
  We study exploration using randomized value functions in Thompson Sampling (TS)-like algorithms in reinforcement learning. We show that when we use 1) a single random seed in each episode, and 2) a Bernstein-type magnitude of noise, we obtain a worst-case $widetildeOleft(HsqrtSATright)$ regret bound for episodic time-inhomogeneous Decision Process.
  arXiv  Detail & Related papers  (2021-02-19T01:42:50Z)
• Nearly Optimal Regret for Learning Adversarial MDPs with Linear Function Approximation [92.3161051419884]
  We study the reinforcement learning for finite-horizon episodic Markov decision processes with adversarial reward and full information feedback. We show that it can achieve $tildeO(dHsqrtT)$ regret, where $H$ is the length of the episode. We also prove a matching lower bound of $tildeOmega(dHsqrtT)$ up to logarithmic factors.
  arXiv  Detail & Related papers  (2021-02-17T18:54:08Z)
• Reward-Free Exploration for Reinforcement Learning [82.3300753751066]
  We propose a new "reward-free RL" framework to isolate the challenges of exploration. We give an efficient algorithm that conducts $tildemathcalO(S2Amathrmpoly(H)/epsilon2)$ episodes of exploration. We also give a nearly-matching $Omega(S2AH2/epsilon2)$ lower bound, demonstrating the near-optimality of our algorithm in this setting.
  arXiv  Detail & Related papers  (2020-02-07T14:03:38Z)
• Frequentist Regret Bounds for Randomized Least-Squares Value Iteration [94.47472987987805]
  We consider the exploration-exploitation dilemma in finite-horizon reinforcement learning (RL) In this paper, we introduce an optimistically-perturbed variant of the popular randomized least-squares value (RLSVI) Under the assumption that the Markov decision process has low-rank transition dynamics, we prove that the frequentist regret of RLSVI is upper-bounded by $widetilde O(d2 H2 sqrtT)$ where $ d $ are the feature dimension, $ H $ is the horizon, and $ T $ is the total number of
  arXiv  Detail & Related papers  (2019-11-01T19:48:57Z)

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.