IMPRS course 'Machine Learning - Zero to Hero'

1.  General information

Date: October 6-10, 2025 from 09:00 - 17:00
Location: MPI-BGC, B0.002
Teacher: Christian Reimers
Category: Skill course
1.0 CP for the whole course

The course is targeted to all interested PhD candidates. No pre-knowledge or experience in Machine Learning is required. Basic programming skills in Python are recommended.

2.  Agenda

09:00-10:00 Introduction to the Course
10:00-10:55 Lecture Block 1

• Loss functions
  
• Linear Regression
• K nearest neighbours (KNN)

10:55-11:10 Break
11:10-12:00 Lecture Block 1 (cont.)
12:00-13:00 Break
13:00-17:00 Practical 1

June 3: Why this already solves every problem

09:00-10:00 Recap of Practical Task Day 1
10:00-10:55 Lecture Block 2

• Kernel trick for linear regression
• KNN with a changed similarity measure
• Decision Tree
• Random Forests

10:55-11:10 Break
11:10-12:00 Lecture Block 2 (cont.)
12:00-13:00 Break
13:00-17:00 Practical 2

June 4: Neural Networks

09:00-10:00 Recap of Practical Task Day 2
10:00-10:55 Lecture Block 3

• Feed Forward Neural Networks
• Stochastic Gradient Decent
• Convolutional Neural Networks
• RNN / LSTM
• Transformer

10:55-11:10 Break
11:10-12:00 Lecture Block 3 (cont.)
12:00-13:00 Break
13:00-17:00 Practical 3

June 5: Dimensionality Reduction

09:00-10:00 Recap of Practical Task Day 3
10:00-10:55 Lecture Block 4

• Principle Component Analysis (+ t-distributed stochastic neighbor Embedding)
• Auto Encoder
• Foundation Models

10:55-11:10 Break
11:10-12:00 Lecture Block 4 (cont.)
12:00-13:00 Break
13:00-17:00 Practical 4

June 6: Probabilistic Prediction

09:00-10:00 Recap of Practical Task Day 4
10:00-10:55 Lecture Block 5

• Pinball Loss
• VAE
• Diffusion Models

10:55-11:10 Break
11:10-12:00 Lecture Block 5 (cont.)
12:00-13:00 Break
13:00-17:00 Practical 5

3.  Registration

Please register here.