Machine Learning 1 (Winter Term 2026/2027)

Overview

• Course consisting of:
  • Lectures (2 hours/week) in TRE/PHYS/E (Zellescher Weg 16) on Fridays, 09:20–10:50
  • Exercise group sessions (2 hours/week):
    • POT/0081/H, Tuesdays, 16:40–18:10, and alternatively:
    • BAR/SCHÖ/E, Thursdays, 14:50–16:40
  • Self-study
  • Final examination
• Language (course and examination): English
• Registration (OPAL). Additional rules for course registration may apply, depending on the study program.
• Lecturer: Bjoern Andres
• Teaching assistants: Emil Powierski, Lucas Fabian Naumann

Contents

• Introduction
• Mathematical foundations
  • Linear and integer optimization for machine learning
  • Partial optimality for machine learning
• Supervised learning
  • Introduction
  • Learning of binary decision trees
    • Hardness
    • Local search algorithm
  • Learning of linear functions
    • Logistic regression
    • Convexity
    • Gradient descent algorithm
    • Support vector machines
  • Learning of composite functions (deep learning)
    • Artificial neural networks
    • Non-convexity
    • Forward propagation algorithm and backward propagation algorithm
    • Attention
    • Transformers
• Unsupervised learning and transductive learning
  • Introduction
  • Clustering
    • Clique partition problem
    • Hardness
    • Local search algorithms
    • LP relaxation, cutting planes
    • Partial optimality
  • Ordering
    • Linear ordering problem
    • Hardness
    • Local search algorithms
    • LP relaxation, cutting planes
    • Partial optimality
  • Classifying
    • Non-binary classification problem
• Supervised structured learning
  • Introduction
  • Graphical models
  • Message passing algorithms
  • Local search algorithms
  • LP relaxation, cutting planes
  • Partial optimality