Learning Action-Transferable Policy with Action Embedding
Abstract
Transfer learning (TL) is a promising way to improve the sample efficiency of reinforcement learning. However, how to efficiently transfer knowledge across tasks with different state-action spaces is investigated at an early stage. Most previous studies only addressed the inconsistency across different state spaces by learning a common feature space, without considering that similar actions in different action spaces of related tasks share similar semantics. In this paper, we propose a method to learning action embeddings by leveraging this idea, and a framework that learns both state embeddings and action embeddings to transfer policy across tasks with different state and action spaces. Our experimental results on various tasks show that the proposed method can not only learn informative action embeddings but accelerate policy learning.
Cite
@article{arxiv.1909.02291,
title = {Learning Action-Transferable Policy with Action Embedding},
author = {Yu Chen and Yingfeng Chen and Zhipeng Hu and Tianpei Yang and Changjie Fan and Yang Yu and Jianye Hao},
journal= {arXiv preprint arXiv:1909.02291},
year = {2021}
}