Feb 9, 2022 · A recent work, called Goal-Conditioned Supervised Learning (GCSL), provides a new learning framework by iteratively relabeling and imitating self-generated ...
Jan 28, 2022 · We revisit GCSL's theoretical foundation and present a simple but effective algorithm for offline goal-conditioned RL via weighted supervised learning.
WGCSL is a simple but effective algorithm for offline goal-conditioned Reinforcement Learning via weighted supervised learning.
This paper studies the problem of constrained offline GCRL and proposes a new method called Recovery-based Supervised Learning (RbSL), which outperforms the ...
In this paper, we revisit the theoretical property of GCSL -- optimizing a lower bound of the goal reaching objective, and extend GCSL as a novel offline goal- ...
Rethinking Goal-conditioned Supervised Learning and Its Connection to Offline RL. Rui Yang1, Yiming Lu1, Wenzhe Li1, Hao Sun2, Meng Fang3, Yali Du4, Xiu Li1 ...
Rethinking Goal-conditioned Supervised Learning and Its Connection to Offline RL. R Yang, Y Lu, W Li, H Sun, M Fang, Y Du, X Li, L Han, C Zhang. International ...
People also ask
Is offline RL supervised learning?
What is the difference between RL and supervised learning?
What is the goal of reinforcement learning RL and how does it work?
What are the differences in the goals of supervised learning versus unsupervised learning?
Abstract—Offline goal-conditioned reinforcement learning. (GCRL) aims at solving goal-reaching tasks with sparse re- wards from an offline dataset.
Rethinking Goal-conditioned Supervised Learning and Its Connection to Offline RL · Rui YangYiming Lu +6 authors. Chongjie Zhang. Computer Science. ICLR. 2022.
Offline reinforcement learning (RL) aims to infer sequential decision policies using only offline datasets. This is a particularly difficult setup, ...