[Lecture] Advanced Machine Learning
In the Advanced Machine Learning lecture we will cover various advanced concepts, techniques, and algorithms. We will place focus both on the mathematical and theoretical aspects, as well as the practical aspects which involve implementing, training, and optimizing machine learning models using real-world datasets. The lecture is organized in four parts. In the first part we will tackle machine learning for graph data including generative models, ranking, and graph neural networks. In the second part we will cover modern generative models such as variational autoencoders, normalizing flows and generative adversarial networks. In the third part we will cover robustness including both attacks on machine learning models (evasion, poisoning) and defenses (certificates). Finally, in the last part we will cover uncertainty quantification techniques such as Bayesian neural networks, Gaussian processes and conformal prediction.
[Lecture] Scientific Machine Learning
Scientific Machine Learning is a new research area that is developing as a field of its own, in which techniques of scientific computing and machine learning are combined and further developed. This results in hybrid methods that are applied to the discretization of partial differential equations, the development of fast and robust solvers, and new modeling techniques.
Keywords are
- Domain-aware scientific machine learning,
- Mathematics-informed machine learning,
- Interpretable scientific machine learning,
- Machine learning-enhanced simulations,
- Hybrid modeling (machine learning + first principle modeling).
[Seminar] Adversarial Machine Learning
In the Adversarial Machine Learning seminar, we will explore the robustness of machine learning models. This seminar will have a practical focus where the students will be split into two sets of teams. One set of teams will focus on developing various attacks to break or manipulate machine learning models, e.g. given an image of a cat design algorithms to add impercetible adversarial noise to the input to cause the model to misclassify it as a dog. The other set of teams will focus on defending against such attacks. The attacks and defenses will be carried out in multiple rounds allowing the attackers and defenders to learn from each other to improve their approach.
[Seminar] Machine Learning methods to predict individual differences with Python
This course will start with learning the basics of programming in Python, so no prior programming knowledge is needed, but helpful. Next will be an introduction to data handling and visualization (i.e., with Seaborn, matplotlib). After that, we will start with basic concepts used in machine learning (i.e., learning, prediction) and helpful methodologies (i.e., cross-validation to prevent overfitting). In the final part of the course, we will learn to apply several machine-learning procedures (i.e., regressions, tree models, deep-learning) and to use model explainers to allow the interpretation of results from otherwise black-box models. As software tools, we will use Jupyther/Colab notebooks. Miniconda or Anaconda should be installed, but all software is freely available. Depending on student demand, this course can be held in German.
[Seminar] Methods of Mathematical Modeling in Life Sciences
The seminar will discuss recent work on applications of mathematical modeling in the life sciences. The focus is on current developments of machine learning and artificial intelligence methods for industrial problems in the fields of pharmaceuticals and agricultural sciences. This seminar discusses different aspects, such as the mathematical methodology behind each method, its computational complexity, and possible applications. In individual cases, publicly available methods will also be applied and the results discussed.
[Seminar] Theory of Machine Learning
In this seminar, the theoretical and algorithmic basics of machine learning are covered. Individual chapters from the book Shai Shalev-Shwartz, Shai Ben-David, Understanding Machine Learning: From Theory to Algorithms, Cambridge University Press, are discussed.
[Seminar] Programming Competition: Modeling of Climate Change Scenarios
In this module you take part in a programming competition about the modeling of risks of climate change. A regression function fed with public weather data needs to be implemented. Likely, the output will be publicly available mortality data. If other reliable data is available, it will be used as well.
The participants can additionally take part with their contribution in a competition sponsored by a business consultancy. The details will be made available at the beginning of the semester; the participation in the competition is voluntary, and it does not require an additional programming effort. The participants' works for the seminar are scored independently of the competition.
Expected skills:
• Good programming skills in Python, Java, or R. (There won't be enough time to acquire these skills during the semester.)
• Interest in the topic of the programming competition
• Work independently to process public data and experiment with various Machine Learning models
• Interest in Data Science
• Helpful are first experiences with Machine Learning models such as neural networks and decision trees.
The goal is to model the impact of climate changes (particularly the temperature) on human health and mortality.
[Seminar] Selected Topics in Convex Optimization
Potential topics include convex optimization for computer-assisted proofs and algorithms, as well as machine learning techniques for quantifier elimination in real closed fields. Prerequisites for this seminar are the lectures "Polynomial Optimization" or "Convex Optimization".
[Seminar] Seminar for Teachers at Grammar and Comprehensive Schools: Practical Algorithms for Instruction
The seminar is targeted at student teachers who are interested in a realistic, youth-oriented teaching structure for the high school level through algorithms in the context of different applications like MP3, JPEG, RSA, GPS, the computation of the Page Rank by search engines etc.
In line with the topic of the Year of Science 2019 (an initiative of the Federal Ministry of Education and Research in Germany), algorithms from the area of artificial intelligence (AI) and machine learning (ML) are treated. Quantum computing and quantum algorithms are another possible focus for the seminar.
For the algorithms and mathematical models, teaching modules are supposed to be created that can supplement the current curricula. The lectures will present the corresponding mathematical basics and a suitable didactic concept.
[Seminar] Inference and Machine Learning for Scientific Research
This seminar shall give an introduction to modern methods of data analysis and machine learning for scientific research. We will focus on methods beyond “black box models” that allow for scientific insight and discoveries.
Contents of the course:
• Fundamentals of Data Analysis and Machine Learning
• Bayesian Statistics and Data Analysis
• Interpretable Machine Learning
• Causal Inference and Modelling
• Inference of Nonlinear Dynamic Systems and Networks