Dynamic Model

Dynamic models aim to represent the evolution of systems over time, focusing on accurately predicting future states based on current observations and underlying physical principles. Current research emphasizes hybrid approaches combining data-driven methods (like neural networks, including Physics-Informed Neural Networks and Graph Neural Networks) with physics-based models, often employing techniques such as Koopman theory and differentiable programming for improved efficiency and accuracy. This work is crucial for diverse applications, including robotics (control, planning, and sim-to-real transfer), autonomous driving, and healthcare (e.g., sepsis treatment and Alzheimer's diagnosis), where accurate prediction of complex, dynamic systems is essential.