SS-MARL combines graph-neural-network message passing with a trust-region constrained update and a new multi-constraint recovery step, reporting better reward and safety trade-offs than MACPO, MAPPO, and InforMARL in cooperative navigation with up to 96 agents.
Safe rlhf: Safe reinforcement learning from human feedback
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.MA 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Scalable Safe Multi-Agent Reinforcement Learning for Multi-Agent System
SS-MARL combines graph-neural-network message passing with a trust-region constrained update and a new multi-constraint recovery step, reporting better reward and safety trade-offs than MACPO, MAPPO, and InforMARL in cooperative navigation with up to 96 agents.