|国家预印本平台
首页|LobsDICE: Offline Learning from Observation via Stationary Distribution Correction Estimation

LobsDICE: Offline Learning from Observation via Stationary Distribution Correction Estimation

LobsDICE: Offline Learning from Observation via Stationary Distribution Correction Estimation

来源:Arxiv_logoArxiv
英文摘要

We consider the problem of learning from observation (LfO), in which the agent aims to mimic the expert's behavior from the state-only demonstrations by experts. We additionally assume that the agent cannot interact with the environment but has access to the action-labeled transition data collected by some agents with unknown qualities. This offline setting for LfO is appealing in many real-world scenarios where the ground-truth expert actions are inaccessible and the arbitrary environment interactions are costly or risky. In this paper, we present LobsDICE, an offline LfO algorithm that learns to imitate the expert policy via optimization in the space of stationary distributions. Our algorithm solves a single convex minimization problem, which minimizes the divergence between the two state-transition distributions induced by the expert and the agent policy. Through an extensive set of offline LfO tasks, we show that LobsDICE outperforms strong baseline methods.

Hongseok Yang、Youngsoo Jang、Geon-Hyeong Kim、Jongmin Lee、Kee-Eung Kim

计算技术、计算机技术

Hongseok Yang,Youngsoo Jang,Geon-Hyeong Kim,Jongmin Lee,Kee-Eung Kim.LobsDICE: Offline Learning from Observation via Stationary Distribution Correction Estimation[EB/OL].(2022-02-27)[2025-05-22].https://arxiv.org/abs/2202.13536.点此复制

评论