Sample Efficient Imitation Learning for Continuous Control

By Fumihiro Sasaki, Tetsuya Yohira, Atsuo Kawaguchi. 2018

65 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Learning Self-Correctable Policies and Value Functions from Demonstrations with Negative Sampling (2019, methodology): “, 2018), and bypassing the learning of reward function (Sasaki et al., 2018).”
Imitation Learning via Off-Policy Distribution Matching (2019, background): “Several papers attempt to relax this on-policy requirement and resolve the sample inefﬁciency problem by designing off-policy imitation learning algorithms, which may take advantage of past logged data, usually in the form of a replay buffer (Kostrikov et al., 2019; Sasaki et al., 2019).”
Task-Relevant Adversarial Imitation Learning (2019, background): “Several recent works have focused on improving the sample efﬁciency of GAIL [Blondé and Kalousis, 2018, Sasaki et al., 2018].”
SQIL: Imitation Learning via Reinforcement Learning with Sparse Rewards (2019, methodology): “First, SQIL is derived from sparsity-regularized BC, while the prior methods are derived from an alternative formulation of the IRL objective [Sasaki et al., 2019] and from support estimation methods [Wang et al.”
Goal-conditioned Imitation Learning (2019, methodology): “Originally, the algorithms used to optimize the policy are on-policy methods like Trust Region Policy Optimization [36], but recently there has been a wake of works leveraging the efficiency of off-policy algorithms without loss in stability [37, 38, 39, 40].”
Positive-Unlabeled Reward Learning (2019, background): “Several authors have noted that the data efficiency of GAIL is improved by using off-policy actor-critic learners [31, 11].”
Imitation Learning from Video by Leveraging Proprioception (2019, methodology): “Even more recently, there has been research on methods that seek to improve on GAIL by, e.g., increasing sample efﬁciency [ Kostrikov et al. , 2019; Sasaki et al. , 2019 ] and improving reward representation [ Fu et al. , 2018; Qureshi et al. , 2019].”
Fighting Failures with FIRE: Failure Identification to Reduce Expert Burden in Intervention-Based Learning (2020, background): “In [16], the authors hypothesize that initializing policies with BC will considerably lower sample complexity, but as pointed out in [23], [24], this does not seem to be the case and requires further investigation.”