@article{461d9760c1074dfc8cc1672367b6e2a3,
title = "Modeling agent decision and behavior in the light of data science and artificial intelligence",
abstract = "Agent-based modeling (ABM) has been widely used in numerous disciplines and practice domains, subject to many eulogies and criticisms. This article presents key advances and challenges in agent-based modeling over the last two decades and shows that understanding agents{\textquoteright} behaviors is a major priority for various research fields. We demonstrate that artificial intelligence and data science will likely generate revolutionary impacts for science and technology towards understanding agent decisions and behaviors in complex systems. We propose an innovative approach that leverages reinforcement learning and convolutional neural networks to equip agents with the intelligence of self-learning their behavior rules directly from data. We call for further developments of ABM, especially modeling agent behaviors, in the light of data science and artificial intelligence.",
keywords = "Agent-based modeling, Artificial intelligence, Data science, Machine learning, Modeling agent decisions and actions",
author = "Li An and Volker Grimm and Yu Bai and Abigail Sullivan and Turner, {B. L.} and Nicolas Malleson and Alison Heppenstall and Christian Vincenot and Derek Robinson and Xinyue Ye and Jianguo Liu and Emilie Lindkvist and Wenwu Tang",
note = "Funding Information: We are indebted to financial support from the National Science Foundation (NSF) through the Method, Measure & Statistics and Geography and Spatial Sciences (BCS # 1638446 ) and the Dynamics of Integrated Socio-Environmental Systems programs (BCS 1826839 and DEB 1212183 ). We thank the participants of the ABM 17 Symposium (sponsored by the above NSF grant; http://complexities.org/ABM17/ ) for input and comments. This project has received funding from the European Research Council (ERC) under the European Union{\textquoteright}s Horizon 2020 research and innovation programme (grant agreement No. 757455 ) and through an ESRC /Alan Turing Joint Fellowship ( ES/R007918/1 ). Funding Information: The RL-CNN approach, though promising and exciting, does not imply that AI, machine learning, and data science are not unbiased, nor does it exhaust the potentials that AI and machine learning can contribute to modelling agent behavior. First, we still emphasize the importance of domain knowledge and theory that are obtained elsewhere (Taghikhah et al., 2022). The mechanism specification in Panel D of Fig. 2, if employed as a starting point for RL network (Panel B), reflects this importance. The mechanisms or rules thus derived—for example, cause-effects and feedback loops in many instances—should be subject to continued examination by domain knowledge and theory. Also, as new data become available, the above RL-CNN or other approaches should be continually used to polish or revise existing rules, even establish new rules. Therefore, continuous real-time data collection is important for not only deriving, but also for validating and renewing, such rules. The concept of “Digital Twins” (DT) is based on this idea of updating, in regular intervals, the data underlying a realistic model used for forecasting. This principle is well-known from weather forecast and widely used in industry (Singh et al., 2022), but has also become the basis of large initiatives to support decision making regarding climate, ocean, and biodiversity, such as the Destination Earth program of the European Commission (Nativi et al., 2021).We are indebted to financial support from the National Science Foundation (NSF) through the Method, Measure & Statistics and Geography and Spatial Sciences (BCS #1638446) and the Dynamics of Integrated Socio-Environmental Systems programs (BCS 1826839 and DEB 1212183). We thank the participants of the ABM 17 Symposium (sponsored by the above NSF grant; http://complexities.org/ABM17/) for input and comments. This project has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (grant agreement No. 757455) and through an ESRC/Alan Turing Joint Fellowship (ES/R007918/1). Publisher Copyright: {\textcopyright} 2023 Elsevier Ltd",
year = "2023",
month = aug,
doi = "10.1016/j.envsoft.2023.105713",
language = "English (US)",
volume = "166",
journal = "Environmental Modelling and Software",
issn = "1364-8152",
publisher = "Elsevier BV",
}