Learning and Decision Making in Multi Agent Software Systems

Background

Multi-agent software systems consist of multiple autonomous agents that interact, collaborate, compete, and make decisions in complex environments. Such systems are increasingly important in areas including artificial intelligence, distributed computing, autonomous systems, robotics, and intelligent software applications. A central challenge in multi agent systems is enabling agents to learn from their environment, adapt their behaviour, and make effective decisions while considering the actions and preferences of other agents. Recent advances in machine learning, reinforcement learning, mechanism design, and algorithmic game theory have provided new approaches for modelling and improving agent interactions. However, achieving reliable, efficient, and explainable decision making in dynamic multi agent environments remains a significant challenge.

Goal/Rationale

This symposium aims to bring together researchers working on learning and decision making in multi-agent software systems, with a focus on understanding how autonomous agents can effectively interact and adapt in complex environments. Key challenges include designing learning mechanisms for strategic agents, achieving cooperation among self-interested agents, handling uncertainty and incomplete information, and ensuring robust performance in dynamic settings.

Recent progress in multi agent reinforcement learning, large language model-based agents, distributed optimisation, and game theoretic approaches has opened new opportunities for developing intelligent multi agent systems. However, important questions remain regarding scalability, stability, coordination, fairness, interpretability, and the theoretical foundations of learning and decision making in such systems. The symposium seeks to explore both theoretical and practical advances, bringing together perspectives from artificial intelligence, computer science, economics, and related disciplines.

Scope

We invite contributions on a broad range of topics related to learning and decision making in multi agent software systems, including but not limited to:

• Multi agent reinforcement learning and adaptive agents
• Decision making under uncertainty and incomplete information
• Coordination, cooperation, and competition among agents
• Game theoretic models and mechanisms for multi agent systems
• Learning in strategic environments
• Distributed and decentralised optimisation
• Agent based modelling and simulation
• Large language model based agents and multi agent AI systems
• Trust, robustness, fairness, and explainability in multi agent decision making
• Applications of multi agent systems in real world domains

The symposium welcomes theoretical, algorithmic, and empirical contributions that advance the understanding and development of intelligent multi agent systems.