Timezone: »

 
Poster
Generalizable Episodic Memory for Deep Reinforcement Learning
Hao Hu · Jianing Ye · Guangxiang Zhu · Zhizhou Ren · Chongjie Zhang

Tue Jul 20 09:00 PM -- 11:00 PM (PDT) @ Virtual
in Poster Session 2 »

Episodic memory-based methods can rapidly latch onto past successful strategies by a non-parametric memory and improve sample efficiency of traditional reinforcement learning. However, little effort is put into the continuous domain, where a state is never visited twice, and previous episodic methods fail to efficiently aggregate experience across trajectories. To address this problem, we propose Generalizable Episodic Memory (GEM), which effectively organizes the state-action values of episodic memory in a generalizable manner and supports implicit planning on memorized trajectories. GEM utilizes a double estimator to reduce the overestimation bias induced by value propagation in the planning process. Empirical evaluation shows that our method significantly outperforms existing trajectory-based methods on various MuJoCo continuous control tasks. To further show the general applicability, we evaluate our method on Atari games with discrete action space, which also shows a significant improvement over baseline algorithms.

Keywords: Reinforcement Learning and Planning
[ Slides] 

Author Information

Hao Hu (Tsinghua University)
Jianing Ye (Peking University)
Guangxiang Zhu (Tsinghua University)
Zhizhou Ren (University of Illinois at Urbana-Champaign)
Chongjie Zhang (Tsinghua University)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors