中文
相关论文

相关论文: Tail Distribution of Regret in Optimistic Reinforc…

200 篇论文

In this paper, we study the application of the Thompson sampling (TS) methodology to the stochastic combinatorial multi-armed bandit (CMAB) framework. We first analyze the standard TS algorithm for the general CMAB model when the outcome…

机器学习 · 计算机科学 2022-06-22 Siwei Wang , Wei Chen

Much of the literature on optimal design of bandit algorithms is based on minimization of expected regret. It is well known that designs that are optimal over certain exponential families can achieve expected regret that grows…

机器学习 · 计算机科学 2024-11-14 Lin Fan , Peter W. Glynn

A stochastic combinatorial semi-bandit is an online learning problem where at each step a learning agent chooses a subset of ground items subject to constraints, and then observes stochastic weights of these items and receives their sum as…

机器学习 · 计算机科学 2017-06-08 Branislav Kveton , Zheng Wen , Azin Ashkan , Csaba Szepesvari

This paper studies regret minimization with randomized value functions in reinforcement learning. In tabular finite-horizon Markov Decision Processes, we introduce a clipping variant of one classical Thompson Sampling (TS)-like algorithm,…

机器学习 · 计算机科学 2021-11-10 Priyank Agrawal , Jinglin Chen , Nan Jiang

We consider reinforcement learning (RL) in episodic Markov decision processes (MDPs) with linear function approximation under drifting environment. Specifically, both the reward and state transition functions can evolve over time but their…

机器学习 · 计算机科学 2024-04-16 Huozhi Zhou , Jinglin Chen , Lav R. Varshney , Ashish Jagmohan

Switching costs, which capture the costs for changing policies, are regarded as a critical metric in reinforcement learning (RL), in addition to the standard metric of losses (or rewards). However, existing studies on switching costs (with…

机器学习 · 计算机科学 2023-02-10 Ming Shi , Yingbin Liang , Ness Shroff

We propose UCBMQ, Upper Confidence Bound Momentum Q-learning, a new algorithm for reinforcement learning in tabular and possibly stage-dependent, episodic Markov decision process. UCBMQ is based on Q-learning where we add a momentum term…

机器学习 · 统计学 2022-03-21 Pierre Menard , Omar Darwiche Domingues , Xuedong Shang , Michal Valko

We study learning algorithms for the classical Markovian bandit problem with discount. We explain how to adapt PSRL [24] and UCRL2 [2] to exploit the problem structure. These variants are called MB-PSRL and MB-UCRL2. While the regret bound…

机器学习 · 计算机科学 2022-05-04 Nicolas Gast , Bruno Gaujal , Kimang Khun

We present the first regret bound for classical online Q-learning in infinite-horizon discounted Markov decision processes (MDPs), without relying on optimism or bonus terms. We first analyze Boltzmann Q-learning with decaying temperature…

机器学习 · 计算机科学 2026-05-18 Rahul Singh , Siddharth Chandak , Eric Moulines , Vivek S. Borkar , Nicholas Bambos

Reinforcement learning (RL) with linear function approximation has received increasing attention recently. However, existing work has focused on obtaining $\sqrt{T}$-type regret bound, where $T$ is the number of interactions with the MDP.…

机器学习 · 计算机科学 2021-02-19 Jiafan He , Dongruo Zhou , Quanquan Gu

Reinforcement learning algorithms are usually stated without theoretical guarantees regarding their performance. Recently, Jin, Yang, Wang, and Jordan (COLT 2020) showed a polynomial-time reinforcement learning algorithm (namely, LSVI-UCB)…

机器学习 · 计算机科学 2024-11-19 Philips George John , Arnab Bhattacharyya , Silviu Maniu , Dimitrios Myrisiotis , Zhenan Wu

In this paper, the problem of distributed optimization is studied via a network of agents. Each agent only has access to a stochastic gradient of its own objective function in the previous time, and can communicate with its neighbors via a…

最优化与控制 · 数学 2024-01-29 Yuchen Yang , Kaihong Lu , Long Wang

We consider the problem of controlling an unknown linear quadratic Gaussian (LQG) system consisting of multiple subsystems connected over a network. Our goal is to minimize and quantify the regret (i.e. loss in performance) of our strategy…

系统与控制 · 电气工程与系统科学 2021-08-19 Sagar Sudhakara , Aditya Mahajan , Ashutosh Nayyar , Yi Ouyang

We study algorithms using randomized value functions for exploration in reinforcement learning. This type of algorithms enjoys appealing empirical performance. We show that when we use 1) a single random seed in each episode, and 2) a…

机器学习 · 计算机科学 2022-10-14 Zhihan Xiong , Ruoqi Shen , Qiwen Cui , Maryam Fazel , Simon S. Du

Contextual bandits are a rich model for sequential decision making given side information, with important applications, e.g., in recommender systems. We propose novel algorithms for contextual bandits harnessing neural networks to…

机器学习 · 统计学 2022-03-01 Parnian Kassraie , Andreas Krause

We consider the problem where M agents collaboratively interact with an instance of a stochastic K-armed contextual bandit, where K>>M. The goal of the agents is to simultaneously minimize the cumulative regret over all the agents over a…

机器学习 · 计算机科学 2022-11-16 Jiabin Lin , Shana Moothedath

We study the Non-Stationary Reinforcement Learning (RL) under distribution shifts in both finite-horizon episodic and infinite-horizon discounted Markov Decision Processes (MDPs). In the finite-horizon case, the transition functions may…

机器学习 · 计算机科学 2026-03-31 Ha Manh Bui , Felix Parker , Kimia Ghobadi , Anqi Liu

This paper is in the field of stochastic Multi-Armed Bandits (MABs), i.e. those sequential selection techniques able to learn online using only the feedback given by the chosen option (a.k.a. $arm$). We study a particular case of the rested…

机器学习 · 统计学 2024-11-28 Marco Fiandri , Alberto Maria Metelli , Francesco Trov`o

The principle of optimism in the face of uncertainty is prevalent throughout sequential decision making problems such as multi-armed bandits and reinforcement learning (RL). To be successful, an optimistic RL algorithm must over-estimate…

机器学习 · 计算机科学 2021-12-07 Aldo Pacchiano , Philip J. Ball , Jack Parker-Holder , Krzysztof Choromanski , Stephen Roberts

We study learning in periodic Markov Decision Process (MDP), a special type of non-stationary MDP where both the state transition probabilities and reward functions vary periodically, under the average reward maximization setting. We…

机器学习 · 计算机科学 2023-03-20 Ayush Aniket , Arpan Chattopadhyay