BriefGPT.xyz
Dec, 2022
通过分叉智能体来提高强化学习的泛化性能
Improving generalization in reinforcement learning through forked agents
HTML
PDF
Olivier Moulin, Vincent Francois-Lavet, Mark Hoogendoorn
TL;DR
本研究提出了受深度神经网络初始化和迁移学习启发的不同初始化技术，探讨其对生态系统方法的速度和效果的影响。
Abstract
An
eco-system
of
agents
each having their own policy with some, but limited, generalizability has proven to be a reliable approach to increase
ge
→