Teaching · Spring 2025

AI for Science Seminar

Undergraduate seminar · Spring Semester 2025 · ETH Zürich

ETH course catalogue Moodle materials Presentation schedule

Course code252-5256-00L

TermSpring 2025

LevelUndergraduate seminar

Kick-off19 Feb 2025

Presentations24 May 2025

Overview

Artificial intelligence and machine learning offer significant potential to revolutionize the fundamentals of scientific computation and discovery today. The goal of this seminar course is to expose students to recent developments in AI for Science.

Students work in pairs, each pair presenting and discussing a recent paper at the end-of-semester presentation session.

Course information

Lecturers: Prof. Dr. Niao He (OAT Y21.1), Dr. Zebang Shen (OAT Y21.2), Dr. Ya-Ping Hsieh (OAT Y19)
Kick-off meeting: 19 February 2025, 12:15–14:00, CAB G 56
Presentation sessions: 24 May 2025, 9:00–16:00, OAT Seminar Room
Language: English
Materials: Announcements and materials are distributed through ETH Moodle. The course catalogue is the authoritative record.

Schedule of presentations (FS2025)

9:00 - 9:25Score-Based Generative Modeling through Stochastic Differential Equations

Presenters: Oliver Pitsch, Fabio Sutter
Supervisor: Dr. Ya-Ping Hsieh

9:25 - 9:50Adjoint Matching: Fine-tuning Flow and Diffusion Generative Models with Memoryless Stochastic Optimal Control

Presenters: Johannes Lorenz, Haroldas Plytnikas
Supervisor: Riccardo De Santi

9:50 - 10:15Attention Is All You Need

Presenters: Lassi Luukkonen, Gabriel Lodi
Supervisor: Riccardo De Santi

10:15 - 10:35Coffee Break

10:35 - 11:00E (n) equivariant graph neural networks

Presenters: Tianxiang Hu, Kristjian Zafirovski
Supervisor: Artur Goldman

11:00 - 11:25Highly accurate protein structure prediction with AlphaFold

Presenters: Pascal Feigenwinter, Oliver Sieberling
Supervisor: Artur Goldman

11:25 - 11:50Accurate structure prediction of biomolecular interactions with AlphaFold 3

Presenters: Renas Sahin, Peer Rheinboldt
Supervisor: Dr. Zebang Shen

11:50 - 13:30Lunch

13:30 - 13:55DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking.

Presenters: Valentin Rossier, Tim Keuning
Supervisor: Dr. Ya-Ping Hsieh

13:55 - 14:20Evolutionary-scale prediction of atomic-level protein structure with a language model

Presenters: Asma Ahmad, Romy Viehweg
Supervisor: Dr. Zebang Shen

14:20 - 14:45MatterGen: a generative model for inorganic materials design

Presenters: Christopher Golling, Luc Tanner
Supervisor: Dr. Zebang Shen

Past presentations (FS2024)

Group 1Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations