It is the preparation of the paper submission for DHM, HCI International 2023. The title is Policy-Based Reinforcement Learning for Assortative Matching in Human Behavior Modeling.
Abstract
Human behavior is the potential and expressive capacity (mental, physical, and social) of human individuals or groups to respond to internal and external stimuli. We explore assortative matching as a typical human behavior in virtual networked communities. We propose a modeling approach based on MAS(Multi-Agent System) and policy-based reinforcement learning to simulate human behavior through various environmental parameter settings and agent action strategies. In our experiment, reinforcement learning serves specific agents who learn from the environment status and competitor behaviors, then optimize strategy to achieve better results. This work simulates both the individual and group level, showing some possible paths for forming relative competitive advantages. This modeling approach can help further analyze the evolutionary dynamics of human behavior, communities, and organizations on various socioeconomic topics.
Keywords
Reinforcement learning, Game theory, Multiagent system, Human behavior modeling
Submitted on arXiv.org. Review it here.