AI for Science Seminar (FS2025)
Artificial intelligence (AI) and machine learning (ML) offer significant potential to revolutionize the fundamentals of scientific computation and discovery today. The goal of this seminar course is to expose student to the recent development of "AI for Science".
Lecturers: Prof. Niao He (OAT Y21.1), Dr. Zebang Shen (OAT Y21.2), Dr. Ya-Ping Hsieh (OAT Y19) .
Kick-off Meeting: February 19, 2025; 12:15-14:00; CAB G 56
Presentation Sessions:
May 24, 9:00-16:00 (OAT Seminar Room),
Useful Links: Course Catalogue (252-5256-00L). All materials are available at Moodle including annoucements and Q&A.
Schedule of Presentations
| Time | Presenter(s) | Paper Title | Supervisor |
|---|---|---|---|
| 9:00 - 9:25 | Oliver Pitsch Fabio Sutter |
Score-Based Generative Modeling through Stochastic Differential Equations | Dr. Ya-Ping Hsieh |
| 9:25 - 9:50 | Johannes Lorenz Haroldas Plytnikas |
Adjoint Matching: Fine-tuning Flow and Diffusion Generative Models with Memoryless Stochastic Optimal Control | Riccardo De Santi |
| 9:50 - 10:15 | Lassi Luukkonen Gabriel Lodi |
Attention Is All You Need | Riccardo De Santi |
| 10:15 - 10:35 | |
Coffee Break | |
| 10:35 - 11:00 | Tianxiang Hu Kristjian Zafirovski |
E (n) equivariant graph neural networks | Artur Goldman |
| 11:00 - 11:25 | Pascal Feigenwinter Oliver Sieberling |
Highly accurate protein structure prediction with AlphaFold | Artur Goldman |
| 11:25 - 11:50 | Renas Sahin Peer Rheinboldt |
Accurate structure prediction of biomolecular interactions with AlphaFold 3 | Dr. Zebang Shen |
| 11:50 - 13:30 | |
Lunch | |
| 13:30 - 13:55 | Valentin Rossier Tim Keuning |
DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking. | Dr. Ya-Ping Hsieh |
| 13:55 - 14:20 | Asma Ahmad Romy Viehweg |
Evolutionary-scale prediction of atomic-level protein structure with a language model | Dr. Zebang Shen |
| 14:20 - 14:45 | Christopher Golling Luc Tanner |
MatterGen: a generative model for inorganic materials design | Dr. Zebang Shen |
Past Presentations (FS2024)
| Group ID | Presenter(s) | Paper Title | Presentation Slides |
|---|---|---|---|
| 1 | Joel Reimann Timon Wattenhofer |
Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations | file |
| 2 | Sonja Joost Gohar Tamrazyan |
B-PINNs: Bayesian physics-informed neural networks for forward and inverse PDE problems with noisy data. | file |
| 3 | Luca Apolloni Eren Homburg |
Fourier Neural Operator for Parametric Partial Differential Equations. | file |
| 4 | Simon Storf |
Choose a transformer: Fourier or galerkin. | file |
| 5 | Khaled Kerouch Jinwei Zhang |
The Deep Ritz Method: A Deep Learning-Based Numerical Algorithm for Solving Variational Problems. | file |
| 6 | Marcel Maciejczyk Franz Schwinn |
Solving high-dimensional partial differential equations using deep learning. | file |
| 7 | Yifei Han Jonathan Seele |
Understanding and mitigating gradient flow pathologies in physics-informed neural networks. | file |
| 8 | Nicola Witzig Andrin Zoller |
Self-consistent velocity matching of probability flows | file |
| 9 | Lino Hofstetter Damiano Meier |
A Tutorial on the Non-Asymptotic Theory of System Identification | file |
| 10 | Zhiang Chen Longxiang Jiao |
Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations. | file |
| 11 | Gaspard Krief Emilia Pucher |
Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators | file |