Improved Regret Bounds for Undiscounted Continuous Reinforcement Learning

Kailasam Lakshmanan 1 Ronald Ortner 1 Daniil Ryabko 2
2 SEQUEL - Sequential Learning
Inria Lille - Nord Europe, CRIStAL - Centre de Recherche en Informatique, Signal et Automatique de Lille (CRIStAL) - UMR 9189
Abstract : We consider the problem of undiscounted reinforcement learning in continuous state space. Regret bounds in this setting usually hold under various assumptions on the structure of the reward and transition function. Under the assumption that the rewards and transition probabilities are Lipschitz, for 1-dimensional state space a regret bound of $\tilde{O}(T^{\frac{3}{4}})$ after any $T$ steps has been given by \citet{contrl}. Here we improve upon this result by using non-parametric kernel density estimation for estimating the transition probability distributions, and obtain regret bounds that depend on the smoothness of the transition probability distributions. In particular, under the assumption that the transition probability functions are smoothly differentiable, the regret bound is shown to be $\tilde{O}(T^{\frac{2}{3}})$ asymptotically for reinforcement learning in 1-dimensional state space. Finally, we also derive improved regret bounds for higher dimensional state space.
Document type :
Conference papers
Complete list of metadatas

https://hal.inria.fr/hal-01165966
Contributor : Daniil Ryabko <>
Submitted on : Sunday, June 21, 2015 - 6:57:00 AM
Last modification on : Friday, March 22, 2019 - 1:35:16 AM

Identifiers

  • HAL Id : hal-01165966, version 1

Citation

Kailasam Lakshmanan, Ronald Ortner, Daniil Ryabko. Improved Regret Bounds for Undiscounted Continuous Reinforcement Learning. International Conference on Machine Learning (ICML), Jul 2015, Lille, France. ⟨hal-01165966⟩

Share

Metrics

Record views

3687