Introduction to machine learning


Prof. Dr. Jochen J. Steil


Institut für Robotik und Prozessinformatik
Mühlenpfordtstraße 23 Telefon: 391 - 7451


Di. 8:00 - 9:30 Uhr (online webinar)
Fr. 8:00 - 9:30 Uhr (online webinar)

Start: 21.04. StudIP subscription is open
Please check the announcement in the StudIP on paricipation, tools, etc.
Note: the lecture is given in english language.

Klausurtermin: Montag, 03.08.2020






Rania Rayyes


Studierende der Informatik, IST und verwandter Gebiete im 6. Semester Bachelor oder im Master.

With successful completion of the module, the students possess the following knowledge and capabilities. They are able to

  1. understand and correctly apply basic concepts of machine learning
  2. analyse and formalize a machine learning problem
  3. distinguish between typical machine learning methods
  4. select a suitable method for a learning problem
  5. compare and judge machine learning methods wrt their capacity
  6. implement machine learning methods and apply them practically
  7. apply and parametrise respective tools
  8. judge strength and weaknesses of machine learning in applications
  9. recognize ethical issues in the application of machine learning

The modul is complementary, but useful for preparation of the master module "Mustererkennung" at the IFN.

It is advised not to take it before the 4. semester, see also "preconditions".


The lecture deals with applied mathematics. The course assumes knowledge in mathematics as acquired in the introductory course in mathematics in the computer science curriculum. Some knowledge in statistics is useful. Concepts used are in particular (conditional) probabilities, the exponential function (in multiple dimensions), basic linear algebra and multidimensional differential calculus (differentiation, chain rule, etc.)


Fundamental principles and theories of machine learning und the underlying mathematical and statistical methods are introduced and learning problems are formalized. Important fundamental terminology, concepts and methods are treated, in particular for regression, among those are

  1. model selection, machine learning bias vs. parameter optimization
  2. training, test and validation
  3. generalization, overfitting, regularization
  4. linear regression, generalized linear models
  5. non-linear models, neural networks
  6. classification
  7. estimatimation, unbiased minimal variance estimators
  8. concept learning, decision trees, random forests
  9. methods of lazy learning
  10. unsupervised learning
  11. Gaussian mixtures, Gaussian mixture regression
  12. Unified Regression Model


Bishop, 2016, Pattern Recognition & Machine Learning
Mitchell, 1997, Machine Learning,
Barber, 2012, Bayesian Reasoning and Machine Learning (online version)
online Kurs:

weiteres wird in der Vorlesung nach Bedarf bekanntgegeben


Unterlagen werden über StudIP/Moodle zum download zur Verfügung gestellt.


1 Prüfungsleistung: mündliche Prüfung (20-30 Minuten) oder eine Klausur (90 Minuten), die Prüfungsform wird in der Vorlesung bekanntgegeben.
Eine Studienleistung in Form der Bearbeitung des online Kurses