392139 Introduction to Machine Learning (int. Track) (V) (WiSe 2026/2027)

The lecture will introduce basic techniques in machine learning, in particular probability based methods. It starts by discussing methods for classification and subsequently (re)-introducing regression in a Bayesian framework as maximum likelihood and maximum a posteriori estimation and proceeds by regarding parameter estimation as a probabilistic process. It introduces concept learning and some of its most popular and widespread applications, e.g. classifaction of data given in form of list of attributes and decision trees. Further topics are unsupervised problems such as clustering and ethical questions in machine learning.

The course consists of three blocks.

::: 1. Block: Introduction and Classification

- Intro: What is machine learning?

- Classification:
lazy learning, logistic regression, concept learning, naive Bayes

- Modelling pipeline, preprocessing, model selection and evaluation:
feature normalization, cross-validation, hyper-parameter selection,
ROC curves

::: 2. Block: Regression, Kernels and Feature Selection

- Regression:
error minimization, ridge regression, probabilistic perspective

- Kernels: kernel trick, kernel ridge regression

- Feature Selection:
wrapper, filters, sparse regression

::: 3. Block: Random Forests and Clustering

- Decision Trees and Random Forests

- Clustering:
DBSCAN, hierarchical clustering, k-means

- Outlook: ethical, legal, social issues:
Privacy, Fairness

In this course, the use of AI tools is permitted for tips on programming syntax / writing (yellow tier in techfak regulations). You are required to be transparent about AI use and
explain and defend your work. Use of AI tools beyond this permission may be sanctioned.

Teilnahmevoraussetzungen, notwendige Vorkenntnisse

Good knowledge of mathematics as taught in the first semesters is indispensible.
We also recommend knowledge about basics of probability theory.

The lecture is part of the international track and will be given in English.

Literaturangaben

There will be lecture notes available.