SS25

Qualifikationsziele:

Die Studierenden kennen unterschiedliche Stichprobenverfahren und können je nach Anwendungsfall das passendste Stichprobenverfahren auswählen und praktisch umsetzen.

Enrollment key: sampling

Time and Dates: Thursdays, 12:15 pm to 1.45 pm, M114 (main building). We will have 5 meetings in the first half of the semester: April 24, May 8 and 22 (note the correction!),  June 5 and 12; one meeting for further mutual exchange will take place at the end of the semester (tbd.).

The course is part of the Bachelor’s programmes in Statistics and Data Science (150 major or 60 ECTS minors: `Methoden und Techniken des wissenschaftlichen Arbeitens’,  P16.1, and  WP 12.1 / WP 13.1, respectively).  Bachelor's students from the "old POs" and Master’s students are most welcome as well. A certificate can be issued for active personal attendance.

Minor students and Bachelor's degree Erasmus students who will not attend a seminar can obtain 3 ECTS via an examination/small project at the end of the course.

Enrollment key: ResTech25

Enrolment key: ml25

Key: dcqd2025

Dates:  28.7-8.8.2025

Termine:

• Der Einschreibeschlüssel lautet: "linalg2025"

Schedule:

• The enrolment key is: "DAS25"

• Bachelor Statistik und Data Science: WP7 Ausgewählte Gebiete der statistischen Modellierung
• Nebenfach Statistik und Data Science (60 ECTS): WP6 Einführung in die Bayes-Statistik
• Nebenfach Statistik und Data Science (30 ECTS): WP8 Einführung in die Bayes-Statistik
• Bachelor Informatik mit integriertem Anwendungsfach Statistik: WP40 Einführung in die Bayes-Statistik
• Bachelor Wirtschaftsmathematik: P17 Ausgewählte Gebiete der angewandten Statistik
• Andere Studiengänge auf Anfrage

Welcome to this introductory course on causality! This is where we gather all the material and organize the in-person sessions.

Dates [Summer Term 2025]

Lecture: Monday, 12h15 - 13h45
Lecture/Exercise (alternating): Tuesday, 08h30 - 10h00


If self-enrollment in this moodle course does not work automatically, you need to register via the backdoor ;)
--> Enrollment key: backdoor

password: savage

Times and Dates: Monday, 2.15 to 3.45 pm in A119, and Tuesday, 5.30 to 7.00 pm in F007. The lectures start on 29.04.

Contents: Decision theory deals with rational decisions under uncertainty. It has high interdisciplinary importance, for example, in the analysis and support of decisions in business administration or finance (e.g. investment strategies), economics or sociology (rational choice theory), medicine (e.g. expert systems) or engineering (e.g. autonomous control). Moreover, statistical decision theory can be seen as a formal framework for choosing analysis methods (optimal tests or estimators, best classification algorithms, etc.). This general view, understanding statistics and machine learning as special cases of decision theory, plays a fundamental role in the critical analysis and problem-adequate generalization of any data-based learning procedure.

The course first discusses the general structure of decision problems, including fundamental decision principles. Then it analyzes and characterizes the Bayes and minimax criteria as extreme poles to deal with (state) uncertainty and develops modern alternatives in the context of complex uncertainty (ambiguity).

Curricular Embedding: 6 ECTS, Master's programme in Statistics and Data Science: "Narrow electice (2 out of 3)" in the Methodology and Modelling track, "wider elective" in the Social Statistics and Data Science and the Econometrics tracks, "general elective" for all tracks; elective in the Master's programme Versicherungs- und Finanzmathematik; options for other programmes upon request. 

In the seminar "Foundation Models in e-Health" we will delve into the rapidly evolving landscape of AI, focusing on the practical use and implementation of foundation models for e-health applications. These models, which serve as the backbone for a wide range of AI applications, offer powerful tools for tackling complex problems across various domains. Throughout the seminar, we'll explore key techniques for fine-tuning and adapting these models to specific e-health tasks, ensuring participants gain a solid understanding of how to leverage their capabilities effectively.

The seminar is designed to bridge the gap between theory and practice, providing master’s students with hands-on experience and actionable insights. We will examine the real-world challenges and do case studies in medical domain.

Recommended prerequisites: Deep learning; Python, PyTorch, TensorFlow, We would also hold the seminar in English and also allow students from other courses (especially DS students)

Organization:

• Enroll key: s25_seminar_ehealth
  
• First meeting: Friday, 11:00-12:30 am,  April. 18th at Ludwig str. 33, Seminar Room (144)
• Weekly meeting:  Friday, 11:00-12:30 am

• [Zoom](link)

• A first meeting will occur at the beginning of the semester (scheduled in agreement with the participants), during which the seminar topics will be briefly introduced and assigned to the participants.
• Seminar type: Block-type and in-person
• Language: The seminar will be held in English

Enrolment key: RegCor25

Format: In-person lectures and lab sessions

Dates & Rooms:
Wed 10-12 @ Main Building A 017
Thu 14-16 @ Schellingstr. 3 (S) - S 001

Grading: Oral exams at the end of the semester.

Der Schlüssel zum Einschreiben ist Stat2NFSoz25.

Format

Class

• Time: Thursday 12:15 - 13:45 h.
• Location: B 106.

Enrollment key

• The enrollment key is: "I2ML".

Short Description: You will learn how to tackle practical challenges in ML using R. Topics include an introduction to an ML toolbox in R, (advanced) tuning techniques, custom ML pipelines, classifier calibration, handling class imbalances, ML benchmark experiments with hypothesis testing, ML model ensembling/stacking, and ML interpretability. Sessions consist of brief methodological overviews followed by hands-on exercises, requiring laptops for full engagement.

Prerequisite: To participate in this lecture, you must have completed and passed the courses:

• Introduction to Machine Learning (I2ML)
• Programming with Statistical Software (R), ideally with a good grade as good knowledge of R is required

Kick-off: Wednesday, April, 23rd from 09:00 - 12:00 (s.t.) (check Moodle for up-to-date information)

Enrolment key: appml25

The course is an introduction to widely used algorithms and data structures, focusing on those most useful in data science. Students are introduced to how data is represented in memory, from individual characters and numbers to complex structures representing hierarchies and graphs. The standard analytical tools used to determine an algorithm's efficiency and the accuracy of its results are also covered.

This course can be taken as an elective module "WP1" and is recommended for the 6th semester (4th semester if you are motivated) in the Bachelor's program. Note that this course can not be credited towards the Master's degree. It targets students with an interest in programming, computer science, and algorithmic thinking. If you liked ProgR, we think you will also like this course.

Schedule:

Description:

In the rapidly evolving field of biomedical research, generating large-scale high-dimensional datasets has become commonplace. This, in turn, has motivated the development and application of new statistical and machine-learning methods to gain valuable insights from such data. In this seminar, students will learn about the most recent advances in data science and statistics aimed at analyzing large-scale biomedical data. The topics covered include (but are not limited to): Compositional data analysis, Variable selection, Network inference and analysis, Self-supervised deep learning, Selective inference, High-dimensional clustering, etc.

Organization:

• A first meeting will occur at the beginning of the semester (scheduled in agreement with the participants), where the seminar topics are briefly introduced and assigned to the participants.
• Seminar type: Block-type and in-person
• Language: The seminar will be held in English
• Enrollment to the seminar will be done by the lecturer after seminar assignment
  

Instructors:

• Lecture: Every Monday 16:00 - 18:00 (c.t.)  -> Room: Geschw.-Scholl-Pl. 1 (M) / M 105
• Labs: Every Thursday 10:00 - 12:00 (c.t.) -> Room: Geschw.-Scholl-Pl. 1 (M) - M 014

• Note: Due to labor day (public holiday), 
  • the first lab session will be on May 8
  • the first lecture session will be on May 5

• Singularity

The field of Automated Machine Learning is thriving, and researchers make tremendous progress in democratizing machine learning. An important new model class is the class of foundation models, and we will discuss how AutoML techniques can be applied to such models, and what these models offer for AutoML. Of course, the seminar will also cover other recent advances, such as human-in-the-loop AutoML.

More concretely, we will discuss topics along the lines of:

1. How can we use AutoML to tune the hyperparameters of foundation models?
2. Can we use foundation models in AutoML?
3. How can we incorporate human experts in the AutoML process?

The following papers are samples of the seminar content:

1. https://www.nature.com/articles/s41586-024-08328-6
   
2. https://proceedings.neurips.cc/paper_files/paper/2022/hash/cf6501108fced72ee5c47e2151c4e153-Abstract-Conference.html
   
3. https://openreview.net/forum?id=VTTL6x0z6V
   
4. https://arxiv.org/abs/2412.07820

Statistik II für Statistiker: Wahrscheinlichkeitstheoretische Grundlagen der Statistik

Termine

• Wiederholungsübung: tbd

Einschreibeschlüssel

• Der Einschreibeschlüssel lautet "stats25wdh".