BriefGPT.xyz
Jan, 2019
离散化连续动作空间的策略优化
Discretizing Continuous Action Space for On-Policy Optimization
HTML
PDF
Yunhao Tang, Shipra Agrawal
TL;DR
本文研究了对连续控制中动作空间的离散化对于基于策略优化的影响,发现动作空间的离散化采用可分解动作分布的策略可以有效地解决离散动作数量的爆炸性增长,并且在复杂动态高维任务上可以通过在策略中使用序数参数化引入自然排序从而获得性能显著提升的优越表现。
Abstract
In this work, we show that discretizing action space for
continuous control
is a simple yet powerful technique for
on-policy optimization
. The explosion in the number of discrete actions can be efficiently addres
→