Solving hard-exploration environments in an important challenge in
reinforcement learning. Several approaches have been proposed and studied, such
as intrinsic motivation, co-evolution of agents and tasks, and mu
本文探讨了如何在多智能体环境下,运用扩展后的 Deep Q-Learning Network,使两个由独立的 Deep Q-Networks 控制的 agents,相互作用以玩经典的电子游戏乒乓球,以及通过改变 Pong 经典奖励方案,演示出竞争和合作性行为的出现。研究表明 Deep Q-Networks 可以成为在高度复杂环境中研究分散式学习的多智能体系统的实用工具。