The existence of complex molecules are a substantial part of the difference between Planet Earth and Big Chunks of Rock in Space. Understanding and accurately predicting molecular properties is a major objective in science but also in analytics, pharmaceuticals, renewables, etc. From the machine learning point of view, these problems are not only relevant but also methodologically interesting given the structural complexity of molecules and the combinatorial size of chemical (search) space.
In this seminar, we will study fundamental molecular representations and suitable machine learning approaches:
- string-based:
- smiles and alternatives; language models
- graph-based:
- message passing networks
- geometric:
- equivariant architectures
- ML in density functional theory
- alphafold: predicting conformations
- generation by diffusion models
Prerequisites
Participants should have acquired a working knowledge of machine learning (at least one set of lectures plus exercises).
Eligibility for your degree
This is a "MSc Pflichtseminar" in the Physics MSc program. Participants are expected to give a presentation and prepare a written report to earn 6CP. Students from other courses or programs who do not want to write a report can earn 2CP by giving a presentation only. Participants are required to attend all presentations.
Registering
Just drop in for an overview of the topics on Monday, April 17th, 2023 at 14h15 in Seminar room 11 of Mathematikon, INF 205.
Future time slots will be agreed amongst the participants.