Core Francisco Park

Ph.D. Candidate at Harvard Physics

prof_pic.jpg

corefranciscopark@g.harvard.edu

Cambridge, MA, USA

Blog and life not ready yet

Hi! I’m Core Francisco Park (pronounced Corae), I am a 5th year graduate student at Harvard Physics.

I work on developping robust, reliable and deployable ML tools and methods for the physical sciences. I define robust, reliable and deployable in a more specific way than “good” or “well performing”:

  • Robustness: The model is expected to work when some assumptions of the data are not met. Out-of-distribution generalization in method-space.
  • Reliable: Different sanity checks on the model has been done, and one can estimate the undertainty from the outputs.
  • Deployable: The model ends up getting used and is not just a project you would archive after publishing.

Of course, these are goals which I think through and try to achieve.

Given these principles, I worked on different real data problems with ML and statistics, ranging from studies of a brain smaller than a grain of salt to cosmic volumes and from very theoretical studies to very practical studies.

I do things outside of work: see the life tab :blush:.

Feel free to ask me anything!

-Core

latest news

Mar 20, 2024 Our follow up paper of our project vdm4cdm_2d is on arxiv.
Mar 07, 2024 I gave a talk about our project vdm4cdm_2d at the ITC Luncheon: “Debiasing with Diffusion: Probabilistic reconstruction of Dark Matter field from galaxies”
Jan 23, 2024 I gave a quick presentation of our project vdm4cdm_3d on using Diffusion Models to generate dark matter fields at the workshop “AI-driven Discovery in Physics and Astrophysics” at Kavli IPMU.
Dec 09, 2023 I will be presenting two works at NeurIPS 2023 Workshops:
  1. Our work on using Diffusion Models to generate Dark Matter maps conditioned on stellar mass fields will be in the Workshop on Machine Learning and the Physical Sciences.
  2. Our work on correcting shadowed spectra in hyperspectral images using Iterative Logistic Regression and Bayesian Inference will be in the Workshop on Tackling Climate Change with Machine Learning.
Dec 05, 2023 Our work tracking individual neurons in C.elegans using synthetically augmented data is published in Nature Methods.