IMPRS / ODSL ML Block Course

16 Mar 2026, 09:00 → 26 Mar 2026, 17:00 Europe/Berlin

Jed Homer (LMU/ODSL), Nicole Hartman (TUM)

The Origins Data Science Lab (ODSL) is organizing a block course on "Introduction to Machine Learning" targeted towards PhD students in physics. Interested MSc students from LMU and TUM are welcome as well.

The courses will take place at the Garching campus, the room for each session is noted on the agenda.

The two-week introductory block course (16.03.2025 - 27.03.2025) will cover the following topics:

• First Week:
  • Supervised learning  and inference
  • Machine Learning theory (bias / variance trade-off)
  • Intro to Neural Networks (universal approximation theorem, optimizers)
  • Deep learning frameworks (e.g, pytorch)
  • Deep Generative Models (Normalizing flows, flow matching, and VAEs)
• Second Week:
  
  • Convolutional Neural Networks
  • Deep sets
  • Transformers / vision transformers
  • Clustering algorithms 
    
    • Classical (k-means / PCA)
    • Neural (Maskformers)
  • Building bigger: neural scaling laws
  • Training tips and tricks
    
     

Please register for the Event using the Registration Link.

Lecturers:

• Nicole Hartman, nicole.hartman@tum.de
• Jed Homer

Forms of credit: ECTS points are available (details forthcoming). If you're taking the course for credit, there will be an exam at the end of the course ( 2 weeks after end of lectures ).

Certificates of participation will be provided on request.

Github: https://github.com/odsl-team/block-course-mar26-ML

^ for tutorials and homework exercises.

Monday 16 March

14:00 → 17:00 Session 1: Intro to ML and Supervised Learning

Location: MPP, Alps room (A.1.01/03)

Directions: Main auditorium located on the first floor of the Max Planck Institute for Physics (MPP). After entering the building via its main entrance on the Boltzmannstrasse, please take the stairs to the right of the Heisenberg cafe to get to the first floor. When reaching the first floor, the main auditorium is directly on the right-hand side and facing the Boltzmannstrasse.

Plan:
- Welcome + logistics
- Intro AI, ML, and deep learning
- Machine Learning: Key ideas
- Linear models (regression)
- Classification
- Perceptron
- Multi-class classification Softmax

Bonus: Linear algebra review: https://cs229.stanford.edu/notes2022fall/cs229-linear_algebra_review.pdf

Tuesday 17 March

09:00 → 12:00 Session 2: Intro to NN and optimizers

Location: MPP, A.1.01 Alps Sued

• Softmax
• Limitations of perceptrons
• Intro to NN / MLP
• Universal approx thm
• Pytorch
• Optimizers

14:00 → 17:00 Session 3: Automatic Differentiation

The "guts" behind deep learning

• Types of gradient estimation algorithms
  • finite difference
  • Symbolic regression
• Forward and reverse mode automatic differentiation

Wednesday 18 March

09:00 → 12:00 Lecture 3: Automatic Differentiation Tutorial

• Implementing pullbacks for composition of functions
• NN from scratch
• Implement Adam

14:00 → 17:00 NN training -- tips and tricks

Location: MPP, A.1.01 Alps Sued

• Machine Learning Theory
• Regularization Techniques
• Tips and tricks for debugging networks

Thursday 19 March

09:00 → 12:00 Lecture 4: Clustering + Decision Trees

Location: MPP, A.1.01 Alps Sued

14:00 → 16:00 Tutorial 4: Clustering

PCA

Speakers: Malin Horstmann (TUM), Matthias Vigl (TUM), Richard Fuchs (ODSL)

Monday 23 March

14:00 → 17:00 Session 7: Convolutional Neural Networks

Speaker: Nicole Hartman (TUM)

Tuesday 24 March

09:00 → 12:00 Session 8: Deep Sets

Speaker: Nicole Hartman (TUM)

14:00 → 17:00 Session 9: Transformers

Speaker: Nicole Hartman (TUM)

Wednesday 25 March

09:00 → 12:00 Session 10: Vision transformers

Speaker: Nicole Hartman (TUM)

14:00 → 17:00 Session 11: Learning to cluster

• k-means / VAE
• Maskformers
• Semantic Segmentation / panoptic segmentatation

Speaker: Nicole Hartman (TUM)

Thursday 26 March

09:00 → 12:00 Session 12: Neural scaling laws

Speaker: Nicole Hartman (TUM)