Multifidelity, domain decomposition, and stacking for improving training for physics-informed networks
Join DynamicsAI’s seminar on Friday, April 25, 2025 at 12:00 – 1:00 p.m. Pacific for a hybrid event featuring Dr. Amanda Howard of Pacific Northwest National Laboratory (PNNL)!
Register now and get virtual access info.
About the Seminar: Physics-informed neural networks and operator networks have shown promise for effectively solving equations modeling physical systems. However, these networks can be difficult or impossible to train accurately for some systems of equations. One way to improve training is through the use of a small amount of data, however, such data is expensive to produce. We will introduce our novel multifidelity framework for stacking physics-informed neural networks and operator networks that facilitates training by progressively reducing the errors in our predictions for when no data is available. In stacking networks, we successively build a chain of networks, where the output at one step can act as a low-fidelity input for training the next step, gradually increasing the expressivity of the learned model. We will finally discuss the extension to domain decomposition using the finite basis method, including applications to newly-developed Kolmogorov-Arnold Networks. Applications will be discussed for accelerating simulations of computational fluid dynamics.
About the Speaker: Amanda Howard is a mathematician at Pacific Northwest National Laboratory specializing in computational fluid dynamics and scientific machine learning (ML). She received her PhD in applied mathematics from Brown University in 2018 and her BS in mathematics from Stanford University in 2012. Her PhD research focused on computational methods for fluid-solid multiphase flows. Her current research includes scientific ML methods for cases with limited data, including multifidelity and physics-informed ML.
