About Sci-ML Symposium

The second student-focused scientific machine learning (SciML) symposium at Georgia Tech is dedicated to the development and applications of SciML methodologies led by current students in a wide array of applications. The primary goal of this symposium is to showcase student talents and contributions through the invited and contributed talks. Although this symposium is student-focused, everyone is welcome to attend irrespective of their student status. This event originated from the graduate Special Topics Course on SciML at Georgia Tech. In compliance with FERPA, only the group names (rather than the author names) will be provided in the schedule for the projects that were conducted by the students of the course.

Code of Conduct: The organizational staff of the SciML Symposium is committed to providing a positive symposium experience for all attendees, regardless of gender, gender identity and expression, sexual orientation, disability, physical appearance, body size, race, age, religion, or national and ethnic origin. We encourage respectful and considerate interactions between attendees and do not tolerate harassment of symposium participants in any form. Symposium participants violating these standards may be sanctioned or expelled from the symposium at the discretion of the symposium organizers.

Zoom - Google Calendar Invites

19th Nov: Quick Link to join Zoom Session 1
21st Nov Quick Link to join Zoom Session 2

Dates and Venue

19th Nov, 2024 - Session 1

Click to Join in Zoom!
8:00-8:15AM 2B1G: Optimize Hydraulic Conductivity Estimation using Fourier Neural Operator
8:15-8:30AM CompOpt: Physics-Informed Fourier Neural Operators for Photonic Device Simulation and Optimization
8:30-8:45AM The Natural Disasters: Predicting Ground Temperatures and the Active Layer Thickness for Permafrost
8:45-9:00AM Bio Team: Modeling the FitzHugh-Nagumo System with Scientific Machine Learning
9:00-9:15AM Magdalini Koukouraki - ESPCI (Invited Talk): Experimental Investigation of Water Wave Scattering by a Vertical Plate

21st Nov, 2024 - Session 2

Click to Join in Zoom!
8:00-8:15AM Physics GP: Towards discovery of minimal structure of Physics-Informed Neural Networks
8:15-8:30AM AIRO: Aeroacoustic Noise Predictions Using Physics-Informed Neural Networks
8:30-8:45AM Physics Team 3: Scientific Machine Learning for Cardiac Electrophysiology Simulations
8:45-9:00AM Team 8: EP-PINN Reduced FHN modeling in Detailed Cardiac Potentials
9:00-9:15AM Eric Qu - UC Berkeley (Invited Talk): The Importance of Being Scalable: Improving the Speed and Accuracy of Neural Network Interatomic Potentials Across Chemical Domains

Guest Speakers

Magdalini Koukouraki (she/her) completed her undergraduate studies on fundamental physics at the National and Kapodistrian University of Athens with a specialisation on astrophysics. Then, she pursued a Master on Wave Physics and Acoustics at the Laboratoire d’Acoustique de l’Université du Mans (LAUM) in Le Mans, followed by an internship on the experimental characterization of nonlinear water waves at the Laboratoire de Physique et de Mécanique des Milieux Hétérogènes (PMMH) in Paris. She is currently a PhD student at PMMH, working on the control of water waves using time-varying metamaterials. This work is supervised by CNRS researchers Philippe Petitjeans, Agnès Maurel and Vincent Pagneux.
Eric Qu (he/him) is a second-year Computer Science PhD Student at University of California, Berkeley, Berkeley AI Research Lab (BAIR), advised by Aditi Krishnapriyan. He is also a visiting researcher at Meta Fundamental AI Research (FAIR), FAIR Chemistry Group. He completed his B.Sc. at Duke Kunshan University and Duke University, during which he has worked at Microsoft Research. His primary research interests lie in Geometric Deep Learning and AI for Science, with a current focus on exploring how to scale up machine learning models to tackle complex challenges in scientific domains.

Organizers

Dr. Raphaël Pestourie (he/him) is an Assistant Professor in Scientific Machine Learning at the School of Computational Science and Engineering, Georgia Tech. He was a Postdoctoral Associate at MIT before joining GaTech, and earned his PhD in Applied Mathematics from Harvard University. He leads a research group with the goal to extend the horizon of accurate models for the optimization of engineering solutions. For example, introducing models where trial and error and heuristics are the state of the art for practitioners. His group develops fast approximate PDE models and scientific machine learning models that combine AI models and scientific models, end to end. These new models enable the ressource-efficient and large-scale optimization of engineering solutions in the following areas:
  • Multifidelity, multiscale, and multiphysics problems simulation.
  • Scientific machine learning for optimization.
  • Metamaterials design.
  • Sophie Bekerman (she/her) is a first-year PhD student with a focus on developing machine learning methods to model multiscale problems across biology and physics.  Working in Dr. Raphael Pestourie’s group, her research focuses on creating generalizable approaches that drive scientific discovery.