Lecture Principles of Deep Learning (MA-INF 4118, 6CP)

Geometric deep learning is a subfield of machine learning that focuses on processing and learning from data with geometric structures, such as graphs, meshes, and point clouds. Due to the often high dimensional and complex data and smaller set of training examples in this field, incorporating prior knowledge about the intrinsic structure and symmetries is essential when designing neural networks. When done right, this enables them to excel in tasks like molecular property prediction, drug discovery, and computer vision applications involving 3D data.

Course Details

  • Curriculum: Master Artificial Intelligence, Master Computer Science
  • Registration: Basis
  • Material: eCampus
  • Effort: 2SWS lecture + 2SWS exercise, 6CP
  • Exam: Tue. 23.02.2027, 10:00–12:00 and Wed. 17.03.2027, 12:00–14:00
  • Midterm: Date announced in first lecture.

Participation

The lecture has a participation limit and will require you to apply in Basis before the start of the semester. The details will be announced through official channels in due time. Students in the AI study program are admitted with priority.

Lecture

  • Time: Friday, 12-14 c.t.
  • Place: Wegelerstraße 10, Kleiner Hörsaal Mathematik (1.001)
  • First lecture: 16.10.26

Exercises

Exercise 1

  • Time: Monday, 10-12 c.t.
  • Place: Friedrich-Hirzebruch Allee 6-8 - Room 3.035b
  • First exercise: 26.10.26

Exercise 2

  • Time: Wednesday, 12-14 c.t.
  • Place: Friedrich-Hirzebruch-Allee 5, Hörsaal 7 (1.004)
  • First exercise: 28.10.26

Exercise 3

  • Time: Thursday, 10-12 c.t.
  • Place: Friedrich-Hirzebruch Allee 6-8 - Room 3.035b
  • First exercise: 29.10.26

Exercise 4

  • Time: Thursday, 16-18 c.t.
  • Place: Nußallee 9, Hörsaal X (1.002)
  • First exercise: 29.10.26