Reinforcement Learning
FlowPG: Action-constrained Policy Gradient with Normalizing Flows
Second, learning the flow model requires sampling from the feasible action space, which is also challenging.
JointPolicySearchforMulti-agentCollaboration withImperfectInformation
To learn good joint policies for multi-agent collaboration with imperfect informationremains afundamental challenge.
Model-Free Reinforcement Learning with the Decision-Estimation Coefficient Dylan J. Foster
In what follows, we review the Decision-Estimation Coefficient and Estimation-to-Decisions meta-algorithm (Section 1.2), highlighting opportunities for improvement.
Towards Minimax Optimal Reinforcement Learning in Factored Markov Decision Processes Yi Tian
We study minimax optimal reinforcement learning in episodic factored Markov decision processes (FMDPs), which are MDPs with conditionally independent transition components.
AcceleratingReinforcementLearningthroughGPU AtariEmulation
CuLE is available at https://github.com/NVlabs/cule.
WideningthePipelineinHuman-Guided ReinforcementLearningwithExplanationand Context-AwareDataAugmentation
Human explanation (e.g.,interms offeature importance) hasbeenrecently used to extend the communication channel between human and agent in interactive machinelearning.
