Skip to main content
Seminar | LANS Seminar | Mathematics and Computer Science Division

Scientific Deep Learning: Reproducibility, Interpretability, and Uncertainty Quantification

LANS Seminar

Abstract: From classifying galaxies and detecting gravitational waves to discovering new materials or new particles in high-energy physics colliders, neural networks are transforming the way science is done and accelerating the pace of progress and discovery. However, these networks can make overly confident or incorrect predictions due to overfitting and their inability to correctly assess the uncertainty. We will discuss some key topics in scientific deep learning, such as uncertainty quantification, interpretability, reproducibility and integrating domain knowledge. This presentation will also showcase some of the exciting data science research at the Leadership Compufing Facility.