Automating Construction of Biochemical and Physical Models with Scientific Machine Learning (SciML)

Date: Thursday 25th September 2025

Abstract

Scientific machine learning, the integration of machine learning into scientific computing, gives many ways to help accelerate the process of scientific discovery. In this talk we will focus on how universal differential equations (UDEs) can be used to augment traditional scientific models with neural networks and generate hypotheses for previously unknown scientific laws through a data-driven process. We will start with the basics, showing how to reliably and reproducibly use the techniques with open source software to generate relativistic corrections to black hole dynamics and discover missing interactions in ecological systems. Then go deeper into the mathematical discussions of how advanced use cases, from building domain-specific architectures to improve learning by directly imposing conservation laws to the mathematical details in the adjoint methods of the training process, in order to give a numerical analysis background on the current state-of-the-art and the challenges that remain for making SciML robust for real-world scientific applications.

More about Dr Chris Rackauckas: 

For his work in mechanistic machine learning, his work is credited for the 15,000x acceleration of NASA Launch Services simulations and recently demonstrated a 60x-570x acceleration over Modelica tools in HVAC simulation, earning Chris the US Air Force Artificial Intelligence Accelerator Scientific Excellence Award. See more at https://chrisrackauckas.com/. He is the lead developer of the Pumas project and received a top presentation award at every ACoP from 2019-2021 for improving methods for uncertainty quantification, automated GPU acceleration of nonlinear mixed effects modeling (NLME), and machine learning assisted construction of NLME models with DeepNLME. For these achievements, Chris received the Emerging Scientist award from ISoP.