Bons-AI: An Agent-Based Model to evaluate the behavior of bonsai grower according to different levels of communication and experience

Guilherme Henrique de Souza Nakahata, Sara Saori Satake, Claus Aranha

Keywords: Agent-Based Modeling, Multi-Agent Systems, Social Learning, Reinforcement Learning, Bonsai

Abstract: Knowledge transfer and social learning are fundamental challenges in multi-agent systems (MAS), particularly in domains where decisions require long term knowledge and environmental factors. In this paper, we introduce Bons-AI, an agent-based model (ABM) that simulates the interaction between bonsai growers with different levels of expertise, aiming to investigate how experience and communication affect the health and style preservation of bonsais. Our model integrates Q-learning with climatic and biological conditions, to simulate plant growth and human decisions. We conducted experiments comparing scenarios with inexperienced growers, autonomous learners, and master–apprentice relationships, the results show that knowledge sharing reduces mortality by 18\% and increases overall health by 9.5\%, highlighting the role of social communication in the learning process. Beyond the specific domain of bonsai cultivation, this work contributes to the MAS by offering a framework for studying adaptive behavior, distinct expertise level, and communication based knowledge transfer in complex environments.

Area: Modelling and Simluation of Societies (SIM)

Generative A I: I acknowledge that I have read and will follow this policy.

Submission Number: 632