Modul

The module will be completed with an oral exam (approx. 30 min).

At the end of the course, students

• know the fundamental principles and problems of machine learning and can relate learning methods to these,
• are able to explain how selected machine learning methods work and can apply these,
• are able to derive and to discuss a statistical analysis of selected learning methods,
• are able to independently develop and apply new learning methods.

none

The course aims for a rigorous and mathematical analysis of some popular machine learning methods with a focus is on statistical aspects. Topics are:

• Regression
  • Empirical risk minimization
  • Lasso
  • Regression trees and Random forests
• Classification
  • Bayes classifier
  • model based classifiers (e.g. logistic regression, discriminant analysis)
  • model-free classifiers (e.g. k nearest neighbors, support vector machines)
• Neural networks
  • training
  • approximation properties
  • statistical analysis
• Unsupervised learning
  • principle component analysis
  • clustering
  • generative models

The modules "Probability Theory" and "Statistics" (M-MATH-103220) are recommended.

Total workload: 240 hours

Attendance: 90 hours

• lectures, problem classes, and examination 

Self-studies: 150 hours

• follow-up and deepening of the course content,
• work on problem sheets,
• literature study and internet research relating to the course content,
• preparation for the module examination