Reinforcement Learning
Regret Bounds for Learning State Representations in Reinforcement Learning
Ronald Ortner, Matteo Pirotta, Alessandro Lazaric, Ronan Fruit, Odalric-Ambrym Maillard
Neural Information Processing Systems http://nips.cc/
State Regularized Policy Optimization on Data with Dynamics Shift
We then demonstrate a lower-bound performance guarantee on policies regularized by the stationary state distribution. In practice, SRPO can be an add-on module to context-based algorithms in both online and offline RL settings.
Importance Resamplingfor Off-policy Prediction
Thoughunbiased, IScanbehigh-variance. Alowervariancealternativeis Weighted IS (WIS). Figure 4: Learning Ratesensitivityplotsinthe Random Walk Markov Chain, withbuffersizen = 15000 andmini-batchsizek = 16.
Neural Temporal-Difference Learning Converges to Global Optima
Qi Cai, Zhuoran Yang, Jason D. Lee, Zhaoran Wang
TD converges at a sublinear rate to the global optimum of the mean-squared projected Bellman error for policy evaluation.
ASPiRe: AdaptiveSkillPriorsforReinforcementLearning
Transferring prior experience to new tasks is central to an agent's adaptability. In this work, we aim to accelerate online reinforcement learning by leveraging prior experience from large offline data.
Distributional Reward Decomposition for Reinforcement Learning
Zichuan Lin, Li Zhao, Derek Yang, Tao Qin, Tie-Yan Liu, Guangwen Yang
Van Seijen et al. [2017] propose to split a state into different sub-states, each with a sub-reward obtained bytraining ageneral valuefunction, andlearnmultiple valuefunctions withsub-rewards. The architecture is rather limited due to requiring prior knowledge of how to split into sub-states.