Reinforcement Learning
MobILE: Model-BasedImitationLearning From ObservationAlone
Weprovide aunified analysis for MobILE, and demonstrate that MobILE enjoys strong performance guarantees for classes of MDP dynamics that satisfy certain well studied notions of structural complexity. We also show that the ILFO problem isstrictly harder than the standard IL problem by presenting an exponential sample complexity separation between ILand ILFO.
MobILE: Model-BasedImitationLearning From ObservationAlone
Weprovide aunified analysis for MobILE, and demonstrate that MobILE enjoys strong performance guarantees for classes of MDP dynamics that satisfy certain well studied notions of structural complexity. We also show that the ILFO problem isstrictly harder than the standard IL problem by presenting an exponential sample complexity separation between ILand ILFO.
Policy Poisoning in Batch Reinforcement Learning and Control
Yuzhe Ma, Xuezhou Zhang, Wen Sun, Jerry Zhu
Neural Information Processing Systems http://nips.cc/
Meta-InverseReinforcementLearningwith ProbabilisticContextVariables
Tothis end,wepropose adeep latent variable model thatiscapable oflearning rewards from demonstrations of distinct but related tasks in an unsupervised way.
Continual
Figure 2:Transfermatrix, see Section 4.2.
Provably Robust Temporal Difference Learning for Heavy-Tailed Rewards
We corroborate our theoretical results with numerical experiments.
