Colby Fronk

Colby Fronk

AI4Science PhD Student

University of California, Santa Barbara

Biography

Colby Fronk is a PhD student studying Scientific Artifical Intelligence and Scientific Computing in the Petzold research lab at UC Santa Barbara. His PhD research lies in the field of neural differential equations, where he develops neural network architectures and numerical methods tailored to neural networks to model scientific processes. Colby has a very broad interest in science and engineering. He works as a data scientist that helps the domain scientist analyze data and develop models for challenging scientific problems.

Interests
  • Artificial Intelligence
  • Neural Differential Equations
  • Multidisciplinary Computational Math, Science, and Engineering
Education

  • PhD in Chemical Engineering, 2024

    University of California, Santa Barbara

  • BSc in Chemical Engineering, 2019

    Rochester Institute of Technology

Experience

 
 
 
 
 
University of California, Santa Barbara
PhD Student Researcher and Teaching Assistant
University of California, Santa Barbara
September 2019 – Present Santa Barbara, California
 
 
 
 
 
Lawrence Livermore National Laboratory
Data Science Summer Institute Program
Lawrence Livermore National Laboratory
June 2023 – September 2023 Livermore, California
While at LLNL, Colby developed AI algorithms for surrogate modeling of plasma physics in nuclear fusion processes. In Dec 2022, LLNL conducted the first successful nuclear fusion experiment and the goal of the research was to improve the models so that nuclear fusion could be scaled up for carbon-free energy production, which the world needs as a solution to climate change.
 
 
 
 
 
Pennsylvania State University
National Science Foundation’s Research Experience for Undergraduates (NSF’s REU) Program
Pennsylvania State University
May 2018 – August 2018 University Park, Pennsylvania
While at Penn State, Colby wrote scientific simulation code to simulate atmospheric cloud formation processes. The research goal was to figure out how cloud droplet formation differed between anthropogenic aerosol particles and secondary aerosol particles released from trees.
 
 
 
 
 
The Hershey Company
Product Commercialization & Process Engineering Co-op
The Hershey Company
May 2017 – May 2018 Hershey, Pennsylvania
While at the Hershey Company, Colby worked with food scientists and engineers to develop new manufacturing processes for new products ranging from snack foods to chocolate bars. This was done through R&D laboratory experiments, pilot plant trials, consumer testing, marketing sample production, and line feasibility tests.
 
 
 
 
 
Johnson & Johnson
Pharmaceutical Development and Manufacturing Sciences Co-op
Johnson & Johnson
June 2016 – January 2017 Spring House, Pennsylvania
While at Johnson & Johnson, Colby worked on a team of 14 research scientists to develop a purification process for an adenovirus-based HIV vaccine.
 
 
 
 
 
Mechanicsburg Area Community Pool
Lifeguard
Mechanicsburg Area Community Pool
May 2013 – September 2015 Mechanicsburg, Pennsylvania

Additional Publications

Quickly discover relevant content by filtering publications.
Colby Fronk, Gwenore Pokrifka, Jerry Harrington, Jose Fuentes (2019). Trees, Aerosols, and Clouds. AMS Annual Meeting.

PDF Cite

Past & Future Talks

Mathematical Model Discovery with Neural Differential Equations
We demonstrate how Neural Differential Equations can be used for data-driven modeling of time-series data and dynamical systems found in science and engineering. We also demonstrate our state-of-the-art symbolic neural ordinary differential equation (neural ODE) tools for the symbolic regression of dynamical systems.
Sep 22, 2023 4:15 PM — 5:30 PM University of California, Santa Barbara
Colby Fronk
Mathematical Model Discovery with Neural Differential Equations
Symbolic Regression for Dynamical Systems with Neural ODEs
We introduce the polynomial neural ODE, which is a deep polynomial neural network inside of the neural ODE framework. We demonstrate the capability of polynomial neural ODEs to predict outside of the training region, as well as perform direct symbolic regression without additional tools such as SINDy.
Apr 21, 2023 11:30 AM — 12:30 PM University of California, Santa Barbara
Colby Fronk
Symbolic Regression for Dynamical Systems with Neural ODEs
Data to Equation, Symbolic Regression with Neural ODEs
We introduce the polynomial neural ODE, which is a deep polynomial neural network inside of the neural ODE framework. We demonstrate the capability of polynomial neural ODEs to predict outside of the training region, as well as perform direct symbolic regression without additional tools such as SINDy.
Aug 25, 2022 Universität Bonn
Colby Fronk
Data to Equation, Symbolic Regression with Neural ODEs
Interpretable Polynomial Neural Ordinary Differential Equations
We introduce the polynomial neural ODE, which is a deep polynomial neural network inside of the neural ODE framework. We demonstrate the capability of polynomial neural ODEs to predict outside of the training region, as well as perform direct symbolic regression without additional tools such as SINDy.
Aug 16, 2022 Schloss Dagstuhl
Colby Fronk
Interpretable Polynomial Neural Ordinary Differential Equations

Projects

*
Eagle Scout Project
An Eagle Scout is the highest rank in the scouting program.
Eagle Scout Project

Contact