Journal article
Multi-Agent Reinforcement Learning-based consensus building for Large-Scale Group Decision Making
Information fusion, v 126(Part B), 103629
Feb 2026
Featured in Collection : Drexel's Newest Publications
Abstract
Large-Scale Group Decision-Making (LSGDM) requires integrating diverse stakeholder opinions in complex, dynamic environments. Traditional GDM methods often lack adaptability to evolving social dynamics and noncooperative behaviors. We propose a Multi-Agent Reinforcement Learning (MARL) framework using Multi-Agent Deep Q-Networks (MADQN) integrated with Social Network Analysis (SNA). Decision-makers (DMs) are clustered into communities, each managed by an agent that autonomously adjusts preferences. A weight penalty mechanism mitigates noncooperative behaviors by reducing the influence of resistant clusters, enhancing consensus efficiency. Simulation results show that MADQN outperforms traditional GDM and other RL-based methods in consensus quality, decision efficiency, and adaptability, reducing iterations and preference adjustments, especially in time-sensitive scenarios.
Metrics
3 Record Views
Details
- Title
- Multi-Agent Reinforcement Learning-based consensus building for Large-Scale Group Decision Making
- Creators
- Yan Tu - Wuhan University of TechnologyWenshuo Wang - Wuhan University of TechnologyZhuang Ma - Tongji UniversityZongmin Li - Sichuan UniversityBenjamin Lev - Drexel University
- Publication Details
- Information fusion, v 126(Part B), 103629
- Publisher
- Elsevier
- Number of pages
- 17
- Grant note
- Hubei Provincial Natural Science Foundation of China: 2025AFB679 The 2025 Theoretical Research Project of Wuhan CPPCC Think Tank: WHZXZK2025B10 National Natural Science Foundation of China: 72174134
This research was supported by the Hubei Provincial Natural Science Foundation of China (grant number 2025AFB679) , the 2025 Theoretical Research Project of Wuhan CPPCC Think Tank (grant number WHZXZK2025B10) , and the National Natural Science Foundation of China (grant number 72174134) .
- Resource Type
- Journal article
- Language
- English
- Academic Unit
- Decision Sciences (and Management Information Systems)
- Web of Science ID
- WOS:001561444200001
- Scopus ID
- 2-s2.0-105013630696
- Other Identifier
- 991022084553704721