With more and more data available for analysis and decision making - from web documents and digital media to sensory data from cameras, microphones, and ubiquitous devices - it becomes increasingly more important to understand how such large volumes of data can be analyzed by computers and used as the basis for new intelligent services, for decision making, and for making computers learn from experience. In companies around the world, from retail and banks all the way to Google, intelligent learning and analysis techniques are used to improve business decisions. Likewise, in science, important discoveries are made easier by automated learning methods, and games and other artifacts are being made adaptive with learning technology.

Within the intelligent systems track of the computer science Master's program, the Algorithms for Data Science course offers in-depth knowledge of different aspects of big data analytics and systems, including algorithmic techniques for analyzing structured and unstructured data that cannot be stored in a single computer because it has enormous size and/or continuously arrives with such a high rate that requires immediate processing. In particular, the topics include classical data mining tasks for massive data and/or data streams, mining massive graphs, and similarity search in massive data.

The lecture series provides an overview of cutting-edge research in triangular AI, which combines a data-intensive approach with human knowledge and context as two additional powerful sources of intelligence. Methods are examined for integrating knowledge from various sources such as previous experience, informed simulations, mathematical descriptions of physical laws, and explicit knowledge provided by humans. The series also explores methods for exploiting the resources of the open, variable contexts in which the system operates, such as active sensing, experimentation and dynamic multimodal interaction with humans. Lectures will also examine several interdisciplinary application areas for triangular AI, as well as its potential for achieving AI that is ethically informed and trustworthy for human users.

Lectures will be given by researchers of the Lamarr Institute for Machine Learning and Artificial Intelligence at the University of Bonn and the Technical University of Dortmund. Lectures will alternate between the two locations, given in presence at one location and live-streamed to the other.

DOWNLOAD PRELIMINARY MEETING SLIDES

(Listed in Basis as: MA-INF 4306 - Lab Development and Application of Data Mining and Learning Systems: Neural ODEs - Praktikum) 

In-context learning, amortized inference, foundation models, zero-shot models and prior fitted networks. These are different labels of the same machine learning paradigm: pre-trained deep neural networks that may be applied in a range of applications, without any further training. LLMs are the most prominent examples of in-context models.

In this lab, we offer several different projects on in-context learning in non-language domains. Concretely, these are project on inference of stochastic processes, function estimation and time series imputation. Their aim is either to understand and extend, or to develop novel in in-context approaches.

All projects include a review of relevant literature and implementations of these ideas. Experience with deep learning and working knowledge of PyTorch or Jax is therefore required. You will report your progress in a series of presentations and a concluding written lab report. 

We will introduce the projects in the preliminary meeting. Topics will be distributed based on preference and interest.