TUBAFdigital - Mathematics of Machine Learning

General Information:

This course is obligatory for the Master studies "Mathematics of Data and Resource Sciences" as well as optional for the Diploma studies  "Angewandte Mathematik" and the Master studies "Geophysik". It takes place each year in winter term. It might be taught in German, if only German speaking students attend the lectures, but the material will be provided in Englisch.

Official course description

Information for SUUUpoRT students:

In the winter term 2025/26 this course will be provided within the SUUUpoRT/TUBAFdigital program. In case SUUUpoRT students want to attend the lectures online, we provide a hybrid mode for the lectures. Please inform the lectuter if the hybrid mode is required. Also there is the possibility to obtain a certificate of participation given the succesful completion of certain quizes. Please enroll in the corresponding learning group "SUUUpoRT-Students"!

Contents:

• statistical learning theory for classification and regression (PAC
  model, empirical risk minimization, Vapnik-Chervonenkis theory)
• linear approaches for classification (perceptron, logistic
  regression, support vector machines, kernel trick)
• feedforward neural networks
• training via stochastic optimization, regularization, validation and testing

Teaching and Dates (WiSe 2025/26)

• 3 SWS Lecture (Prof. Dr. B. Sprungk)
  Mondays, 16:25 - 17:55, room MIB-1113
  Thursdays (even weeks), 16:15 - 17:45, room MIB-1113
• 1 SWS Exercise (M. Sc. Hanyue Gu)
  Tuesdays (even weeks), 08:00 - 09:30, room MIB-1113

Pre-requisites:

Recommended are basic knowledge of  mathematical optimization, numerical analysis and probability theory

Exam modalities:

• The exam is an oral exam of 30 minutes
• The module has 6 credit points.

 Module number: MAML MA Nr. 3694