Past Event: Oden Institute Seminar

In Context Learning for Scientific Computing

Frank Cole, Hedrick Math Fellow, UCLA

Abstract

Transformer-based foundation models, pre-trained on broad datasets, have demonstrated a remarkable ability to adapt to diverse downstream tasks. Their applications span natural language processing, computer vision, and even scientific research. A striking feature of these models is their ability to perform in-context learning: given a prompt containing a few examples of an unseen task, they can make relevant predictions without updating their parameters.  In this talk, I will give an overview of in-context learning and present recent advances in its mathematical foundations. I will focus on the application of transformer neural networks to two representative problems in scientific computing: solution operator learning for PDEs and optimal transport.

Biography

Frank Cole is a Hedrick Math Fellow at UCLA since July 2026. He earned his PhD in mathematics from the University of Minnesota in May 2026. His research interests lie in the mathematics of machine learning and data science and their applications to efficient, reliable scientific computation methods.