Timezone: »

 
Poster
Generalized Off-Policy Actor-Critic
Shangtong Zhang · Wendelin Boehmer · Shimon Whiteson

Tue Dec 10 10:45 AM -- 12:45 PM (PST) @ East Exhibition Hall B + C #188
in Reinforcement Learning and Planning -- Decision and Control »

We propose a new objective, the counterfactual objective, unifying existing objectives for off-policy policy gradient algorithms in the continuing reinforcement learning (RL) setting. Compared to the commonly used excursion objective, which can be misleading about the performance of the target policy when deployed, our new objective better predicts such performance. We prove the Generalized Off-Policy Policy Gradient Theorem to compute the policy gradient of the counterfactual objective and use an emphatic approach to get an unbiased sample from this policy gradient, yielding the Generalized Off-Policy Actor-Critic (Geoff-PAC) algorithm. We demonstrate the merits of Geoff-PAC over existing algorithms in Mujoco robot simulation tasks, the first empirical success of emphatic algorithms in prevailing deep RL benchmarks.

Keywords: Reinforcement Learning and Planning · Reinforcement Learning and Planning -> Decision and Control; Reinforcement Learning and Planning -> Markov Decision Processes;
[ Paper]  [ Slides] 

Author Information

Shangtong Zhang (University of Oxford)
Wendelin Boehmer (University of Oxford)
Shimon Whiteson (University of Oxford)

More from the Same Authors