Exploration, Collaboration, and Applications in Multi-Agent Reinforcement Learning

Research Summary: In recent years, the field of humancentric decision-making has emerged as a critical area of research, driven by its potential to fundamentally reshape how decisions are made across a variety of complex systems. Human-centric decision-making is characterized by considerable variability, which can be both advantageous and problematic. Variability is beneficial in contexts like brainstorming or problem-solving, where diverse perspectives lead to innovation. However, in situations where consistency is crucial- such as public health, safety, and fairness-this variability can have negative consequences. For instance, when radiologists interpret the same mammogram differently, it can result in unnecessary medical procedures. Similarly, in military operations, inconsistent judgments by sonar technicians regarding potential threats can jeopardize both crew safety and mission success. This highlights the critical need for a deeper understanding of decisions where the stakes are high. Thus, it is important to deepen our understanding of the decision-making process in completed and dynamic scenarios that involve multiple decisionmakers. This should guide the creation of validated methods to identify, measure, model, and adjust variability in judgments to improve decision-making in multi-agent environments.