In dieser Projektgruppe werden grundlegende Algorithmen aus den Bereichen Wissensentdeckung und Data Mining erarbeitet und diskutiert. Im Mittelpunkt steht die Erarbeitung der Inhalte für verschiedene Zielgruppen. Die Aufgabe der Studierenden ist die Vermittlung von Wissen in Form eines Coding Nuggets als praktische Anleitung für Fachleute als auch in Form eines Blog Beitrags für interessierte Laien. Dabei wird auf die unterschiedlichen Anforderungen bei der Wissenvermittlung für unterschiedliche Zielgruppen eingegangen.

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, for making computers learn from experience, and for generative artificial intelligence. In companies around the world, from retail and banks all the way to Google, OpenAI, etc., 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. This module is offered as an introduction to Machine Learning for bachelor's students. We will give a practically oriented introduction into the most popular methods from Machine Learning, including regression methods, decision tree methods, clustering, artificial neural networks, kernel methods such as SVM, reinforcement learning, deep learning methods such as, for example,  autoencoders, sequence models, and transformers. This will be complemented with lectures on the most important approaches within computational learning theory. Within the exercises, it is possible to try out the most important methods and popular Machine Learning systems.

Please note that the course will be held in English.

In this seminar, we learn skills and knowledge required for Visual Analytics.

Our goal is to be in a learning process where you explore why these skills are essential through hands-on tasks and active participation

Our targets during the seminar:

• to enable you to find visual representations that facilitate reasoning with data
• to understand computational methods in visualization
• to develop visual interfaces that aid human interpretation of visually represented data, preferably in an interactive manner.

As preparation for our first meeting, please find and bring a couple of real-world examples of visual representations, such as charts, that you find effective in interpreting data. Include a brief summary explaining what you think makes these visualizations useful.

Prerequisite: a basic understanding of data visualization tools such as D3.js (JavaScript) or Plotly (Python) is ideal, though familiarity with Google Spreadsheets is sufficient for participation.

What you will have to do: we plan at least one presentation on your development process and a final report on the visual analytics tools you've created. The course will aim to seamlessly blend theoretical concepts with hands-on practice.

Seminar materials will be accessible on my website, which provides participants with updates, including important dates and announcements.

This lab corresponds to a natural continuation of our seminar last semester on Neural Operators. However, having participated in said seminar is not a requirement. All students are welcome to participate, and no background knowledge on neural operators is required.

Neural Operators employ neural networks to learn mappings between infinite dimensional spaces, which simply means that one can train neural networks to e.g.

1. automatically solve entire families of complex partial differential equations;

2. integrate families of functions; or

3.  infer families of functions and (stochastic) processes from experimental data.

Last seminar dealt with point (ii). In this lab we will explore point (iii), to wit: how neural operators can be leveraged to infer, in the spirit of Bayesian inference, functions from noisy sets of observations. Specifically we will infer Ordinary Differential Equations (ODEs) from noisy observations on their solutions.

Our main goals will be to:

• test these ideas with some simple coding exercises we will do from scratch
• read and discuss about some current issues of the methodology and their possible solutions.

What you will have to do: we will ask you to give a set of presentations about your numerical experiments, and hand in a final report. You will work in groups of up to 3 people.

Requirements: working knowledge of machine learning libraries like PyTorch or Jax

Dates: We will have our preliminary meeting next Monday, April 15, at 10am in Room 3.110 of the B-IT