Seminar: Deep Learning for the Natural Sciences (5 ECTS)
Summer Semester 2024
Organiser: Karnik Ram, Annalena Kofler
Description
Following its success in computer vision and language processing, deep learning is now being increasingly used to augment and accelerate research in the natural sciences. Examples include electro-catalyst discovery for energy storage, fast PDE solvers for weather forecasting, discovery of new anti-biotics, and many other advances that were not possible using traditional methods alone. In this seminar, we will discuss relevant papers in this new area, with an emphasis on the deep learning techniques powering these approaches such as geometric deep learning, generative models, and self-supervised learning.
Format
- Students are expected to study one paper in depth, and present and lead a discussion on it. Apart from the presentation and report, students are also expected to periodically submit one-paragraph summaries of the papers discussed, and participate in the discussions.
- Sessions will be held in-person (with a remote attendance option) once every two weeks, on Tueday afternoon (14:30 - 16:30). There will be two paper presentations in every session. There may also be a catch-up lecture on certain relevant topics from deep learning (eg. diffusion models, graph learning) at the start based on interest.
- All class-related communications are over Discord, and the summaries, presentation, and report are managed on Gradescope.
Prerequisites
A good understanding of machine learning techniques (esp. deep learning), linear algebra, and calculus. Undergraduate students are requested to check with the organizer before enrolling.
Schedule
Location: CIT Seminarraum 00.08.055
Time: 2:30 PM to 4:30 PM
Virtual: Zoom Link