Multi-Agents on a Muti-Layer Network A MADDPG-Based Method
Hyeongtak Yun · 2024
With the continuous improvement of computing ability, reinforcement learning algorithms have been introduced to solve problems in fields such as autonomous driving, robotics, and game theory. Due to the fact that the agents in these fields need to propose effective cooperation or competitive strategies timely, the multi-agent reinforcement learning algorithms are proposed to help agents to face the dynamic and complex environments. In this paper, a multiagent testing environment was constructed to validate the representative algorithm of multi-agent reinforcement learning, MADDPG. And the result showcases its superior performance and stability. These findings are expected to provide valuable insights into enhancing the efficiency of multi-agent configurations.