Research Associate Optimization, Inversion, Machine Learning, and Uncertainty for Complex Systems
Dr. O'Leary-Roseberry received his PhD in 2020 and his MS in 2017 from Computational Science, Engineering and Mathematics at The University of Texas at Austin. His BSE is in Engineering Mechanics and BA in Mathematics are from the University of Wisconsin--Madison in 2015.
Tom's research interests lie at the intersection of computational science, optimization and machine learning. One research focus is in designing efficient and scalable second order optimizers for high dimensional stochastic-nonconvex optimization problems (such as neural network training).
Another research focus is in developing efficient and scalable nonlinear surrogates for expensive "many-query" applications that arise in computational science. These surrogates exploit low dimensional model-based information to design the surrogate models and reduce the amount of training data required. Applications for this work include forward uncertainty quantification of complex climate systems and inverse design for aerodynamic shape optimization.