Mathematics of Machine Learning - WiSe22/23

This lecture will be on mathematical machine learning. We will study just a few concepts and algorithms, and will study in details their mathematical properties. We will consider two main fields:

- Batch learning: in this setting, the dataset is available beforehand. We will study a general franework called empirical risk minimisation, and will see how it applies to the problem of classification. We will in particular study the classical SVM algorithm.

- Sequential learning: in this setting the dataset comes sequentially as an online stream. We will study two main settings here: online learning and bandit theory.