Completion of the module "Machine Learning" or similar prior knowledge in the area of machine learning is highly recommended. Completion of the module "Deep Learning" or the simultaneous enrollment in the module "Deep Learning" is recommended.

The lecture will discuss several fields of machine learning that go beyond the traditional setting of supervised learning, by training predictive models with little or no labeled data.

This includes semi-supervised learning approaches, which exploit unlabeled data; transfer learning approaches, which exploit labeled data from different but related domains; and self-supervised learning approaches that exploit other training signals than the output labels used in standard supervised learning. The lecture will often discuss such approaches based on concrete application problems.