The field of artificial intelligence is witnessing significant developments in multi-agent systems and large language models (LLMs). Researchers are exploring innovative ways to integrate LLMs with evolutionary optimization, hybrid rewards, and enhanced observation to address challenges in multi-agent reinforcement learning. Meanwhile, there is a growing interest in explaining the decision-making and behavior of agents, particularly in complex environments. The use of process mining techniques and novel frameworks is being investigated to provide insights into agent strategies and improve their performance. Noteworthy papers include LERO, which proposes a framework integrating LLMs with evolutionary optimization to address MARL-specific challenges, and AgentNet, which enables LLM-based agents to autonomously evolve their capabilities and collaborate efficiently in a decentralized network. Other notable works, such as the exploration of explainable multi-player MCTS-minimax hybrids and the development of novel attacks on pragmatic multi-agent LLM systems, are also pushing the boundaries of the field.
Advances in Multi-Agent Systems and Large Language Models
Sources
LERO: LLM-driven Evolutionary framework with Hybrid Rewards and Enhanced Observation for Multi-Agent Reinforcement Learning
Exploring Explainable Multi-player MCTS-minimax Hybrids in Board Game Using Process Mining
EncGPT: A Multi-Agent Workflow for Dynamic Encryption Algorithms
An Organizationally-Oriented Approach to Enhancing Explainability and Control in Multi-Agent Reinforcement Learning
Intrinsically-Motivated Humans and Agents in Open-World Exploration
$\textit{Agents Under Siege}$: Breaking Pragmatic Multi-Agent LLM Systems with Optimized Prompt Attacks
AgentNet: Decentralized Evolutionary Coordination for LLM-based Multi-Agent Systems
Foundation Models for Autonomous Driving System: An Initial Roadmap
First Field-Trial Demonstration of L4 Autonomous Optical Network for Distributed AI Training Communication: An LLM-Powered Multi-AI-Agent Solution
Strategize Globally, Adapt Locally: A Multi-Turn Red Teaming Agent with Dual-Level Learning
Interpreting Emergent Planning in Model-Free Reinforcement Learning
LLMs Working in Harmony: A Survey on the Technological Aspects of Building Effective LLM-Based Multi Agent Systems
Advances and Challenges in Foundation Agents: From Brain-Inspired Intelligence to Evolutionary, Collaborative, and Safe Systems
Learning and Improving Backgammon Strategy