M2SE: multiple scenario-semantic exploration for multi-agent cooperative learning

Hong Chen; Liang Sun

doi:10.1117/12.3026818

8 April 2024 M2SE: multiple scenario-semantic exploration for multi-agent cooperative learning

Hong Chen, Liang Sun

Proceedings Volume 13090, International Conference on Computer Application and Information Security (ICCAIS 2023); 130902W (2024) https://doi.org/10.1117/12.3026818
Event: International Conference on Computer Application and Information Security (ICCAIS 2023), 2023, Wuhan, China

Abstract

When applying reinforcement learning to multi-agent decision making, it has been challenging to learn effective strategies in dynamic complex environments. Agents need to consider and interpret the high-level information abstracted from different perspectives of the scenario, which will help them take appropriate actions. However, many existing methods are devoted to learning the end-to-end “observation-action” mapping. In fact, the real-world scenarios usually involve many interacting agents, which makes appropriate end-to-end mapping difficult to obtain. Therefore, this paper proposes a mechanism of multiple scenario-semantic exploration (M2SE) for multi-agent cooperative learning. Based on this mechanism, a multiple scenario-semantic exploration network is proposed to acquire the high-level scenario-semantic information embedded in agents’ observations and to provide agents with heuristics from different views. The M2SE network constructs a “observation-semantic-action” mapping and enables agents to learn more appropriate strategies by fusing the high-level information in the semantic space. It is straightforward to integrate the M2SE into existing MARL algorithms. Experimental findings indicate a notable performance improvement in MARL algorithms when using the M2SE.

(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Hong Chen and Liang Sun "M2SE: multiple scenario-semantic exploration for multi-agent cooperative learning", Proc. SPIE 13090, International Conference on Computer Application and Information Security (ICCAIS 2023), 130902W (8 April 2024); https://doi.org/10.1117/12.3026818

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available