Matt Sampson
Matt Sampson

PhD candidate

about me

I am currently at Princeton University studying ways to advance science through the use of machine learning with a particular focus on modelling dynamical systems and geometric deep learning. Previously I was at the Australian National University where I performed numerical experiments and helped develop a code to simulate cosmic ray propagation through turbulent plasma. I have also worked as a data scientist/computational statistician in fields of information geometry and biology.

cv
Interests
  • machine learning
  • deep learning
  • neural/latent ODEs
  • graph neural networks
  • computational statistics
  • astrophysics
Education
  • PhD astrophysics and machine learning

    Princeton University

  • MA Astrophysics and Machine Learning

    Princeton University

  • BHons Computational Astrohysics

    Australian National University

  • BSc Computational Mathematics

    Queensland University of Technology

  • BSc Physics

    Queensland University of Technology

recent publications
(2024). Path-minimizing Latent ODEs for improved extrapolation and inference. MLST (full version) NeurIPS Workshop for Physical Sciences.
(2024). Disentangling transients and their host galaxies with Scarlet2: A framework to forward model multi-epoch imaging. arXiv preprint arXiv:2409.15427.
(2024). Score-matching neural networks for improved multi-band source separation. Astronomy and Computing. Astronomy and Computing.
(2023). Spotting Hallucinations in Inverse Problems with Data-Driven Priors. Spotlight talk at Machine Learning for Astrophysics. Workshop at the Fortieth International Conference on Machine Learning (ICML 2023), July 29th, Hawaii, USA.
(2023). Turbulent diffusion of streaming cosmic rays in compressible, partially ionized plasma. Monthly Notices of the Royal Astronomical Society 519 (1), 1503-1525.