Professor Yongbing Zhang from the School of Computer Science and Technology at Harbin Institute of Technology, Shenzhen Campus, collaborated with Professor Xiangyang Ji from the Department of Automation at Tsinghua University to achieve significant research progress in the field of AI + diffusion dynamics. Their research findings, titled Reliable Deep Learning in Anomalous Diffusion Against Out-of-Distribution Dynamics, have been published in the international academic journal Nature Computational Science.
As deep learning methods are applied to the identification and characterization of anomalous diffusion, a fundamental challenge has emerged: if the observed trajectories lack the characteristics of the training diffusion models, deep learning methods will lead to incorrect recognition of observed phenomena. This potential risk of erroneous identification hinders the deployment of deep learning methods in real diffusion dynamics research and raises a new question: Can researchers leverage deep learning’s errors to explore potential connections between different diffusion dynamics?
In response to the complex and unknown diffusion dynamics behaviors in real-world scenarios, the research team has proposed, for the first time, a deep learning framework for reliably identifying anomalous diffusion. This framework, driven by AI for Science, aims to transform existing diffusion assessment models and thoroughly discusses the opportunities for deep learning to discover and analyze unknown diffusion patterns from empirical observations. This marks a pioneering step in leveraging AI to deepen human understanding of anomalous diffusion and complex dynamic behaviors.
Paper link: https://www.nature.com/articles/s43588-024-00703-7
