Data Scientist
Data Scientist

Biography

Jonathan Tu is a data scientist at Pacific Northwest National Laboratory (PNNL). Tu earned his undergraduate degrees in aeronautics and astronautics as well as mathematics from the University of Washington. Tu earned his PhD in mechanical and aerospace engineering from Princeton University. At Princeton University, Tu majored in control and dynamical systems and minored in fluid mechanics while performing research work on reduced-order modeling for control of fluid flows. He then moved to University of California Berkeley as a postdoctoral researcher to study the hydrodynamics of microscopic swimmers. Through the course of his training Tu acquired a broad technical background in engineering, applied mathematics, and scientific computing.

Thereafter, Tu worked as a research scientist/engineer at Naval Surface Warfare Center, Carderock Division of the U.S. Navy, leading numerous projects involving machine learning (ML) and other data-driven algorithms. His work included basic research efforts to develop and assess new algorithms, as well as applied work to develop and deploy ML algorithms aboard Navy vessels. His interests include adversarial ML, interpretable ML, and physics-informed ML.

Disciplines and Skills

  • Machine learning
  • Scientific computing
  • Dynamical systems
  • Fluid dynamics
  • Linear algebra
  • Software development

Education

  • PhD in mechanical and aerospace engineering, Princeton University, 2013
  • BS in aeronautics and astronautics, University of Washington, 2008
  • BS in mathematics, University of Washington, 2008

Affiliations and Professional Service

  • Society for Industrial and Applied Mathematics

Publications

2023

  • Brown D.R., C.W. Godfrey, C.A. Nizinski, J.H. Tu, and H.J. Kvinge. 2023. “Robustness of edited neural networks.” In International Conference on Learning Representations (ICLR) 2023 Workshop on Mathematical and Empirical Understanding of Foundation Models (ME-FoMo). PNNL-SA-181779.
  • Brown D.R., C.W. Godfrey, C.A. Nizinski, J.H. Tu, and H.J. Kvinge. 2023. “Robustness properties of edited deep learning models.” In Fortieth International Conference on Machine Learning. PNNL-SA-181441.