장기적 경험을 통해 에이전트들은 협력적인 이웃을 선별하고 선호하는 상호작용 전략을 개발하여, 네트워크 상호성을 높이고 전체적인 협력 수준을 향상시킨다.
LOQA는 상대방의 행동-가치 함수를 모델링하여 상호 협력을 유도하는 분산 강화학습 알고리즘이다.
Multi-agent synchronization tasks (MSTs) require precise timing and coordination among agents to achieve successful outcomes, posing significant challenges for existing multi-agent reinforcement learning (MARL) approaches.
Verco, a novel multi-agent reinforcement learning algorithm, enables agents to generate human-understandable verbal messages to enhance coordination and cooperation.
A novel approach to infer a Group-Aware Coordination Graph (GACG) that captures both agent-pair cooperation and group-level dependencies to facilitate efficient information exchange and decision-making in cooperative multi-agent tasks.
Equivariant network architectures can effectively leverage environmental symmetry to improve zero-shot coordination between independently trained agents in decentralized partially observable Markov decision processes.
The core message of this paper is to introduce a novel approach called Best Response Shaping (BRS) that trains an agent by differentiating through an opponent approximating the best response, in order to learn reciprocity-based cooperative policies in partially competitive multi-agent environments.
Power regularization can mitigate the negative effects of misaligned communication in cooperative multi-agent reinforcement learning (CoMARL) systems by quantifying and controlling the influence (power) that agents exert over each other's decision-making through communication.
This paper proposes an attention-based methodology that integrates domain knowledge into the MARL process by incorporating predefined higher-level tasks, simplifying the learning process and enhancing collaborative behaviors.
ROMA-iQSS enables decentralized agents to independently identify optimal objectives and align their efforts towards a common goal through a combination of state-based value learning and a specialized multi-agent interaction protocol.