Research

At the Data Science chair, we focus on diverse set of research topics, all connected by the common interest of fundamentals in machine learning. Here, we give a general overview about methods and research topics we're currently interested in. For our various applications and projects, click here.

Knowledge Enriched Machine Learning

Leveraging additional domain knowledge for deep learning models, for example knowledge graphs for NLP, additional item attributes in recommendation or knowledge about physical characteristics in Neural PDEs.

Large Language Models

We’re working on adapting Large Language Models such as BERT and BART to solve diverse tasks and work on extracting and including knowledge.

Time Series & Sequence Modeling

Reasonable modeling of time or order is a challenge in many areasm such as sensor data, but also (human) behavior & language. Other examples include RNA, or behaviour sequences in recommendation and many areas more.

Deep Representation and Metric Learning

Developing common approaches for designing and learning representations and metrics for a specific domain or downstream task.

Deep Learning for Imbalanced Data

Improving and developing machine learning techniques which can work with rare events, for example rare events in weather or bee hives, or anomalies in security or ERP data.

Explainable AI

As AI models become increasingly complex, understanding AI decisions becomes increasingly difficult. Here, Explainable AI provides methods to trace decisions and to ensure the appropriate behavior of complex AIs.