Department of Computer Science

Faculty of Mathematics and Science

Eberhard Karls University Tübingen

Module handbook
Machine Learning
Master of Science (M.Sc.)

Released by the Academic Commission of the Department of Computer Science

(updated January 16, 2021)
Contents

Preface 3
Structure and Subject Matter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Credit Points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Types of Courses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Grading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

Master’s Program Machine Learning 5

General Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Subjects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Qualification Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Areas of Studies and Modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Module catalogue for the Master’s degree program Machine Learning 10

Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Legend . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

Study Area: Foundations of Machine Learning 11

ML-4103 Deep Learning (Lecture with tutorials) . . . . . . . . . . . . . . . . . . . . . . . . . . 11
ML-4201 Statistical Machine Learning (Lecture with tutorials) . . . . . . . . . . . . . . . . . . . 13
ML-4202 Probabilistic Inference and Learning (Lecture with tutorials) . . . . . . . . . . . . . . . 14

Study Area: Diverse Topics in Machine Learning 15

Lectures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
ML-4101 Mathematics of Machine Learning (Lecture with tutorials) . . . . . . . . . . . . . . . . 15
ML-4102 Data Literacy (Lecture with tutorials) . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
ML-4301 Numerical Algorithms of Machine Learning (Lecture with tutorials) . . . . . . . . . . . 18
ML-4302 Statistical Learning Theory (Lecture with Tutorials) . . . . . . . . . . . . . . . . . . . 19
ML-4303 Convex and Nonconvex Optimization (Lecture with tutorials) . . . . . . . . . . . . . . . 20
ML-4310 Data Mining and Probabilistic Reasoning (Lecture with tutorials) . . . . . . . . . . . . 21
ML-4320 Time Series (Lecture with tutorials) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
ML-4340 Self-Driving Cars (Lecture with tutorials) . . . . . . . . . . . . . . . . . . . . . . . . . 23
ML-4360 Computer Vision (Lecture with tutorials) . . . . . . . . . . . . . . . . . . . . . . . . . 24
ML-4601 Introduction to Game Theory with Application to Multi-Agent Systems (Lecture) . . . . 25
ML-4350 Reinforcement Learning (Lecture, Tutorial) . . . . . . . . . . . . . . . . . . . . . . . . 27
ML-4410 Neural Data Analysis (Lecture, Tutorial) . . . . . . . . . . . . . . . . . . . . . . . . . 28
INFO-4492 Special Topics in Learning Theory (Lecture, Tutorials) . . . . . . . . . . . . . . . . . 29
ML-4420 Efficient Machine Learning in Hardware (Lecture) . . . . . . . . . . . . . . . . . . . . . 30
Seminars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
ML-4501 Machine Learning Seminar (Seminar) . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
ML-4503 Explainable Machine Learning (Seminar) . . . . . . . . . . . . . . . . . . . . . . . . . 32
INFO-4493 Learning Theory (Seminar ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
ML-4502 Machine learning methods for scientific discovery (Seminar) . . . . . . . . . . . . . . . . 34
Practical Courses and Research Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

1
Module handbook M.Sc. Machine Learning 2

ML-4510 Practical Machine Learning (Practical Course) . . . . . . . . . . . . . . . . . . . . . . 35

ML-4998 Research Project Machine Learning (Independent research project) . . . . . . . . . . . . 36

Study Area: General Computer Science 37

INFO-4315 Advanced Topics in Embedded Systems (Lecture) . . . . . . . . . . . . . . . . . . . . 37
INFO-4194 Behavior and Learning (Lecture, Tutorials) . . . . . . . . . . . . . . . . . . . . . . . 39
INFO-4210 Recurrent and Generative Artificial Neural Networks (Lecture, Tutorials) . . . . . . . 40
INFO-4212 Artificial Neural Networks (Practical Course) . . . . . . . . . . . . . . . . . . . . . . 41
INFO-4213 Advanced Artificial Neural Networks Project (Practical Course) . . . . . . . . . . . . 42
INFO-4214 Cognitive Modeling (Lecture, Tutorials ) . . . . . . . . . . . . . . . . . . . . . . . . 43
INFO-4211 Avatars in Virtual Realities (Practical Course) . . . . . . . . . . . . . . . . . . . . . 44
INFO-4250 Information Processing for Perception and Action (Seminar) . . . . . . . . . . . . . . 45
INFO-4152 Advanced Statistics (Lecture, Tutorials) . . . . . . . . . . . . . . . . . . . . . . . . . 46
INFO-4149 Selected Topics in Database Systems (Lecture, Tutorials) . . . . . . . . . . . . . . . . 47
INFO-4381 Advanced Topics in Human-Computer Interaction (Seminar) . . . . . . . . . . . . . . 48
INFO-4412 Algorithms and Complexity (Lecture, Tutorial) . . . . . . . . . . . . . . . . . . . . . 49
INFO-4241 Programming Languages II (Lecture, Tutorials) . . . . . . . . . . . . . . . . . . . . . 50
INFO-4246 Programming with Dependent Types (Practical Course) . . . . . . . . . . . . . . . . 51
INFO-4248 Interactive Theorem Proving (Lecture, Tutorials) . . . . . . . . . . . . . . . . . . . . 52
BIO-4242 Advanced Java in Bioinformatics (Lecture and tutorials) . . . . . . . . . . . . . . . . . 53
BIO-4311 Microbiome analysis (Lecture, Tutorials ) . . . . . . . . . . . . . . . . . . . . . . . . . 54
BIO4364 Visualization of Biological Data (Lecture, Tutorials) . . . . . . . . . . . . . . . . . . . . 55
BIO-4331 Advances in Computational Transcriptomics (Lecture, Tutorials) . . . . . . . . . . . . . 56
BIO-4210 Practical Transcriptomics (Practical course) . . . . . . . . . . . . . . . . . . . . . . . 58
BIO-4363 RNA Bioinformatics (Seminar) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
MEDZ-4991 Medical Data Science (Lecture, Tutorial) . . . . . . . . . . . . . . . . . . . . . . . . 60

Study Area: Expanded Perspectives 62

ML-5001 Expanded Perspectives (Lecture, Tutorial, Seminars) . . . . . . . . . . . . . . . . . . . 62

Module Master Thesis 63

ML-4999 Master thesis (Independent research work, Master’s thesis (in written form) and oral pre-
sentation) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Preface

Structure and Subject Matter

This handbook describes the modules that make up the Master’s program Machine Learning at the De-
partment of Computer Science (Faculty of Science, Eberhard Karls University Tübingen). The Master’s
program consists of elective-compulsory modules (“Foundations of Machine Learning”), a large variety of
elective modules in the area of machine learning (“Diverse Topics in Machine Learning”) and more computer
science in general (“General Computer Science”), as well as completely free modules (“Expanded perspec-
tives”). Descriptions for the modules and additional areas of studies are given below, containing among
other information the number of credit points required in each. Credit point requirements in a studies area
are fulfilled by completing one or more modules belonging to that area and adding up credit points earned.
Which module belongs to which area(s) can be seen from this handbook’s appendix, the modules list.

Credit Points
Study areas and modules earn crtedit points (also: ECTS points based on the European Credit Transfer
System, or simply credits). Credit points measure a student’s time investment. Following national as well as
international standards (in Germany: Resolution of the Standing Conference of the Minsters of Education
and Cultural Affairs, 24 October 1997), a credit point represents a workload of 30 hours in attended classes
and autonomous study performed by the student. The overall per-semester workload (including nonterm
periods) ought not to exceed 900 hours, resulting in approximately 30 credit points required of the student
each semester. Credit points represent not only time spent physically attending classes, but also time spent
on preparing for and processing classes, as well as autonomous activities such as preparation for exam,
writing the master thesis, and practical projects. Credit points are earned by attending and participating in
courses that make up the modules, and require the completion of course-related tasks.

Types of Courses
Below we detail the general types of courses in the Master’s program (note that some individual courses
might give alternative information in their course descriptions).

Lectures, with and without tutorials In lectures, transfer of knowledge takes the form of a series of
talks by the teacher. Lectures often go hand in hand with tutorials that deepen the students’ understanding
and knowledge about the subject matter and apply the techniques presented in class to concrete examples
and scenarios. Homework commonly accompany this course type. Often, programming and other practi-
cal exercises, in which tasks are completed under direct supervision, form an important part. Typically,
performance measurement and grading is based on a written (or more rarely oral) exam at the end of term.

Practical training are courses in which students finish assigned tasks in small teams, autonomously or
under supervision. Study and exam performance are usually evaluated based on active participation, a
presentation of results and in written reports.

3
Module handbook M.Sc. Machine Learning 4

Research projects are intended to give students an opportunity to get engaged in the ongoing research
conducted in one of the groups and labs participating in this study program, for the duration of one semester.
This course type aims to closely link the Master’s program to current research, and to thoroughly prepare
students for their upcoming Master’s thesis. Study and exam performance are usually evaluated based on
active participation, a presentation of results and in written reports. If applicable, students can participate
in scientific publications.

Seminars are a series of classes in which students take up a specific assigned subject matter and give a
presentation about it in front of their teacher and other co-participants. Usually, handing in a written version
is an additional requirement. Performance is measured and grades assigned based on the presentation, the
written report, and the student’s active participation in class.

Grading
Modules will, as a rule, always be graded. Grades are determined by taking an examination of some sort
– in the case of lectures, this is typically a written test. In certain instances, grading can be based on a
multi-part examination. Details are given in the module descriptions. Grading is performed by the teachers
of individual modules. According to our examination regulations, the grades of each module enter into the
cumulative grade (Master’s degree final grade), weighted by the module’s credit points. An exception are
modules within in ML-EXP (Expanded Perspectives) area of studies: credit points earned here can be used
to fulfill overall credit requirements according to Examination Regulations, §3 (Structure), but grades earned
in this area do not enter into the calculation of the cumulative grade for the Master’s program. This gives
students the possibility to extend their horizons by attending courses that are out of their comfort zone,
without risking a dip in their overall grade.
Master’s Program Machine Learning

General Information
Subjects
The international Master’s Program Machine Learning will enable graduates to analyze, implement, leverage,
and modify techniques of machine learning. As future actors and deciders in the field, they will be compe-
tent in all basic and many advanced areas of machine learning, understanding and suitably applying this
increasingly essential tool for dealing with large datasets, be it in science, industry or alternative domains.
The studies program deals both with generic methods and their applications to specific fields, making it
highly relevant for new career and job market purposes, both in science and industry. Education in problem
solving capabilities is a central training objective.
To pick up on scientific trends and make the best use of the current state of research, the curriculum
relies heavily on the strong research presence on site, in machine learning as well as the wider field of
computer science: top-level researchers in all major methodological branches of machine learning are present
in Tübingen – personnel that will actively engage in teaching for the Master’s Program Machine Learning.
Since the field is obviously very young and currently developing extremely rapidly, training will naturally be
based on the most recent insights and the most pressing research questions of these teaching researchers.
Project work and the Master’s thesis will offer students the opportunity to develop code for research
purposes and their own scientific projects. In this whole Master program, besides professional expertise,
graduates will also acquire language skills and intercultural competence due to the program’s international
nature.

Qualification Objectives
The Master’s Program Machine Learning promotes a focus on research. It expands and deepens methodolog-
ical and technical knowledge, enables graduates to work scientifically, provides the basis for advancing the
field, and prepares graduates for subsequent PhD studies. The program specifically empowers graduates to
take up responsible leading roles and emphasizes a scientific, research-oriented mindset based on independent
thought, judgement and decision-making.
The program explicitly aims to cover the full breadth of the field, ranging from fundamental skills in
mathematics and data handling to advanced methods of data analysis using a variety of methods of machine
learning. We will particular train students to be able to quickly take up new research developments in
the field of machine learning. Alongside aiming for breadth, the program also encourages specialization, in
that modules within one area of studies can be freely combined. In their Master’s thesis, graduates can
take machine learning approaches and methods to tackle a freely chosen area in computer science or an
adjoining field such as bioinformatics or medical informatics. The requisite depth of knowledge to do so will
be obtained due to the program’s consecutive studies plan, which is based on a B.Sc. in computer science
or a neighbouring discipline.

Qualification objectives of this Master’s program are as follows:

Graduates...

5
Module handbook M.Sc. Machine Learning 6

1. ...have further developed the qualifications obtained in their B.Sc. studies in an ongoing process of
academic maturation. They have transferred learned skills to the field of machine learning and gained
facility in applying and implementing technical and non-technical knowledge.

2. ...have obtained expert knowledge in a chosen focus field in the wider area of machine learning. .
3. ...have the necessary breadth as well as depth to quickly acquaint themselves with new developments
in their own area of expertise and its adjacent areas.
4. ...are able to successfully utilize, to critically examine and to further advance machine learning methods
in order to formulate and solve complex problems of research and development in the industry as well
as research.
5. ...have acquired a diverse technical and social skillset (abstraction, analytical and systematic thinking,
teamwork, communication, international and intercultural competence etc.), empowering them to seek
positions of leadership.

6. ...are optimally prepared not only for functions related to research and development, but also for further
responsible and leading positions in the industry or public administration.
Module handbook M.Sc. Machine Learning 7

Areas of Studies and Modules

Foundations of Machine Learning (ML-FOUND): this study area covers the basic, foundational
directions in the field of machine learning that every student is supposed to learn. The modules in this area
are elective modules, and altogether 24 CPs have to be earned in this area.

Diverse Topics in Machine Learning (ML-DIV): this study area contains many different courses of
various aspects of machine learning, ranging from theory, generic methods, implementation details and fields
of applications. Students can choose freely from this area of studies, and thereby set their own focus. All in
all, 36 CPs need to be earned in this area of studies.

General computer science (ML-CS): In this study area students can take part in other courses offered
by the Department of Computer science, for example to broaden their knowledge in a technique they feel
they are still lacking (e.g., databases), or in application domains (e.g., computer vision, bioinformatics).
Students choose courses of a total of 18 CPs.

Expanded Perspectives (ML-EXP): In this study area, students can choose courses freely from almost
all courses (except for sports courses) offered at the University of Tübingen. It is meant to give students the
opportunity to learn about particular application fields (e.g., geoscience, linguistics), improve their language
skills in German (for foreign students) or English (for German students), or learn to reflect upon ethical or
philosophical challenges brought by machine learning. Altogether 12 CPs in this field have to be fulfilled.
Courses taken in this area need to be graded ones, and the grades will show up on the transcript of records,
but the grades will not be taken into account for the cumulative grade of the Master’s program, as stated
above.
Module handbook M.Sc. Machine Learning 8

Structuring and Organizing Your Studies

The Examination Regulations, §3 (Structure), provides details on how to structure the studies in the Master’s
Program Machine Learning over four semesters. Overall, the program requires 120 credit points to be
obtained. More information on modules and types of courses can be found within this module handbook.
Figures 1-3 below show examples of study plans as examples of how one may organize one’s studies; the
ability to freely combine modules within areas of studies ensures that a wide range of studies plans are viable.
For students who plan to spend a semester abroad we recommend to do this in the third semester.

1st. Semester 2nd. Semester 3rd. Semester 4th. Semester

Practical Machine
Deep Learning Statistical Machine Learning
Learning
Numerical Algorithms of
Data Literacy
ML
Probabilistic Inference
Seminar ML
and Learning
Master thesis
Mathematics of ML
Interactive Theorem
Convex and Nonconvex Proving
Optimization
Algorithms and
Complexity
Efficient ML in Hardware Reinforcement Learning

30 LP 30 LP 30 LP 30 LP

ECTS
ML-FOUND Foundations of Machine Learning 24
ML-DIVERSE Diverse Topics of Machine Learning 36
ML-CS General Computer Science 18
ML-EXP Expanded Perspectives 12
Thesis Master thesis 30

Figure 1: Study plan with focus on theory

Module handbook M.Sc. Machine Learning 9

1st. Semester 2nd. Semester 3rd. Semester 4th. Semester

Practical Machine
Deep Learning Statistical Machine Learning
Learning

Data Literacy Self-Driving Cars

Probabilistic Inference
Seminar ML
and Learning
Master thesis
Mathematics of ML
Advanced Java
ML in Graphics and
Vision
Cognitive Modelling Advanced SQL
Ethics in Science
German as Foreign German as Foreign German as Foreign
Language Language Language
30 LP 30 LP 30 LP 30 LP

ECTS
ML-FOUND Foundations of Machine Learning 24
ML-DIVERSE Diverse Topics of Machine Learning 36
ML-CS General Computer Science 18
ML-EXP Expanded Perspectives 12
Thesis Master thesis 30

Figure 2: Study plan with focus on practical (e.g. industrial) applications

1st. Semester 2nd. Semester 3rd. Semester 4th. Semester

Practical Machine
Deep Learning Statistical Machine Learning
Learning

Data Literacy Time Series

Probabilistic Inference
Seminar ML
and Learning
Master thesis
Mathematics of ML Computational
Microbiome Analysis
Neural Data Analysis
Visualisation of
Systems Biology
large-scale data
Ethics in Science
German as Foreign German as Foreign German as Foreign
Language Language Language
30 LP 30 LP 30 LP 30 LP

ECTS
ML-FOUND Foundations of Machine Learning 24
ML-DIVERSE Diverse Topics of Machine Learning 36
ML-CS General Computer Science 18
ML-EXP Expanded Perspectives 12
Thesis Master thesis 30

Figure 3: Study plan with focus on biomedical applications

Module catalogue for the Master’s
degree program Machine Learning

Notes
The following module list specifies the courses offered for the Master Program in Machine Learning, describes
each of them using an abstract of the subject matter, qualification aims, and exam modalities, and associates
them with individual required study areas.

The academic council of the Computer Science Department provides an updated version of the module
catalogue at the beginning of each semester.

Legend

Abbreviations Meaning
Type L = Lecture
S = Seminar
T = Tutorial
P = Practical course
R = Research project
Status c = compulsory
o = optional
CH Credit hours
CP Credit points (= ECTS points)
Type of exam wt = written test
ot = oral test
tp = term paper
op = oral presentation
Duration of exam in minutes
Evaluation g = graded
ug = ungraded (pass / fail)
nt = no test
Calculation of modules possible percentage weighting of grades

10
Study Area: Foundations of Machine
Learning

Module Number: Module title Module

ML-4103 Deep Learning Lecture with tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture with tutorials
Duration 1 semester
Frequency Regularly once a year
Language of English
instruction
Type of Exam Written exam (in case of a small number of participants: oral exam)
Content Within the last decade, deep neural networks have emerged as an indispensable
tool in many areas of artificial intelligence including computer vision, computer
graphics, natural language processing, speech recognition and robotics. This
course will introduce the (practical and theoretical) principles of deep neural
networks and give an overview over the most established training and regular-
ization techniques. The lecture will further discuss the most important network
variants, including convolutional neural networks, generative neural networks,
recurrent neural networks and deep reinforcement learning. Furthermore, the
course will give an overview over the most important architectures (hourglass
networks, skip connections, dense connections, dilated convolutions, permu-
tation invariant networks, siamese networks, etc.). In addition, applications
from various fields will be presented throughout the course. The tutorials will
deepen the understanding of deep neural networks by implementing, training
and applying them using modern deep learning frameworks.
Objectives Students gain an understanding of the theoretical and practical concepts of
deep neural networks including optimization, inference, architectures and ap-
plications. After this course, students should be able to develop and train deep
neural networks, reproduce research results and conduct original research in
this area.

11
Module handbook M.Sc. Machine Learning 12

(still ML-4103)

Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation
Requirement for
Credit Points /

Status
Grade

CH

CP
Lecture L o 2 3 wt/ot 90 g 100
Tutorial T o 2 3
Usability (modules) Foundations of ML;
Requirement for Basic math (linear algebra & analysis, probability and information theory) and
participation coding knowledge (variables, functions, loops, classes, algorithms). Experience
in Python is recommended.
Lecturer Geiger, Zell
Literature Related literature will be listed throughout the lecture.
Module handbook M.Sc. Machine Learning 13

Module Number: Module title Module

ML-4201 Statistical Machine Learning Lecture with tutorials
ECTS 9
Work load
- Contact time Work load Class time Self-Study
- Self study 270 h 90 h / 6 CH 180 h
Lecture type Lecture with tutorials
Duration 1 semester
Frequency Regularly once a year
Language of English
instruction
Type of Exam Written test (in case of a small number of participants: oral tests)
Content The focus of this lecture is on algorithmic and theoretical aspects of statistical
machine learning. We will cover many of the standard algorithms, learn about
the general principles for building good machine learning algorithms, and an-
alyze their theoretical and statistical properties. The following topics will be
covered: Supervised machine learning, for example linear methods; regulariza-
tion; SVMs; kernel methods. Bayesian decision theory, loss functions,
Unsupervised learning problems, for example dimension reduction, kernel PCA,
multi-dimensional scaling, manifold methods; spectral clustering and spectral
graph theory.
Introduction to statistical learning theory: no free lunch theorem; generaliza-
tion bounds; VC dimension; universal consistency;
Evaluation and comparison of machine learning algorithms.
Advanced topics in statistical learning, for example low rank matrix completion,
compressed sensing, ranking, online learning.
Objectives Students get to know the most important classes of statistical machine learn-
ing algorithms. They understand why certain algorithms work well and others
don’t. They can evaluate and compare the results of different learning algo-
rithms. They can model machine learning applications and get a feeling for
common pitfalls. They can judge machine learning algorithms from a theoret-
ical point of view.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 4 6 W 90 g 100
Tutorial T o 2 3
Usability (modules) Foundations of ML;
Requirement for Students need to know the contents of the basic math classes, in particular
participation linear algebra and probability theory.
Lecturer Hein, von Luxburg
Literature The literature for this lecture will be provided at the beginning of the semester.
Module handbook M.Sc. Machine Learning 14

Module Number: Module title Module

ML-4202 Probabilistic Inference and Learning Lecture with tutorials
ECTS 9
Work load
- Contact time Work load Class time Self-Study
- Self study 270 h 90 h / 6 CH 180 h
Lecture type Lecture with tutorials
Duration 1 semester
Frequency Regularly once a year
Language of English
instruction
Type of Exam Written test (in case of a small number of participants: oral tests)
Content Probabilistic inference is a foundation of scientific reasoning, statistics, and
machine learning. The lecture course begins with a general introduction to
basic principles (rules of probability theory, graphical models), then covers the
probabilistic view on many standard settings, like supervised regression and
classification, and unsupervised dimensionality reduction and clustering. In a
parallel thread through the lecture, we will also encounter a number of popu-
lar algorithms for inference in probabilistic models, including exact inference
in Gaussian models, sampling, and free-energy methods. At specific points,
connections and differences to non-probabilistic frameworks will be made.
Apart from mathmatical derivations, the exercises put a focus on practical
programming. In particular, they contain implementations of some content of
the lectures.
Objectives Students gain an intuitive, as well as a mathematical and algorithmic under-
standing of probabilistic reasoning. They acquire a mental toolbox of proba-
bilistic models for various problem classes, along with the algorithms required
for their concrete implementation. Over the course of the lecture, they also
become proficient in the fundamental concept of uncertainty, and the philo-
sophical challenges and pitfalls associated with it. They are empowered to
build, analyse, and use their own probabilistic models for concrete use cases.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 4 6 W 90 g 100
Tutorial T o 2 3
Usability (modules) Foundations of ML; General Computer Science; Expanded Perspectives
Requirement for Standard undergraduate knowledge of mathematics is required, to the extent
participation that is provided, for example, by the course on Mathematics for Machine Learn-
ing (ML 4101).
Lecturer Hennig, Macke
Literature Literature will be listed at the beginning of the semester.
Study Area: Diverse Topics in
Machine Learning

Lectures

Module Number: Module title Module

ML-4101 Mathematics of Machine Learning Lecture with tutorials
ECTS 9
Work load
- Contact time Work load Class time Self-Study
- Self study 270 h 90 h / 6 CH 180 h
Lecture type Lecture with tutorials
Duration 1 semester
Frequency every year
Language of English
instruction
Type of Exam Written exam (in case of a small number of participants: oral exams)
Content The lecture will repeat and introduce basic notions of mathematics used in
machine learning
• Calculus: multivariate calculus (gradient and Hessian), Taylor expan-
sion etc.
• Linear Algebra: eigenvectors, eigenvalues (including variational char-
acterization), singular value decomposition and best low rank approxima-
tion, inverse and pseudo-inverse, norms, basic algorithms and their com-
plexity (solving linear equations, matrix inversion, eigenvectors (power
method)) etc.
• Probability: discrete and continuous probability measures (and mixed
ones), basic notions, generation of random variables, conditional expecta-
tion and independence, law of large numbers and concentration inequal-
ities for rates of convergence, central limit theorem etc.
• Statistics: parametric and non-parametric tests
• Optimization: Lagrangian and dual optimization problem, popular op-
timization techniques and their properties
• Optional: basic functional analysis and approximation theory, curse of
dimensionality

15
Module handbook M.Sc. Machine Learning 16

(still ML-4101)
Objectives Students learn the mathematical foundations for the latter machine learning
courses. In particular,
• they know multivariate calculus and linear algebra as needed in machine
learning lectures
• they can apply probability and statistics and are able to prove basic
properties

• they have an overview of existing optimization techniques and are able

to reformulate equivalent constrained optimization problems

Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation
Requirement for
Credit Points / Status
Grade
CH

CP
Lecture L o 2 3 W 90 b 100
Tutorial T o 2 3
Usability (modules) Diverse Topics in ML;
Requirement for Students need to have basic knowledge in analysis and linear algebra on the
participation level of the bachelor lectures “Mathematik für Informatiker I-III”
Lecturer von Luxburg, Hein
Literature The literature for this lecture will be provided at the beginning of the semester.
Module handbook M.Sc. Machine Learning 17

Module Number: Module title Module

ML-4102 Data Literacy Lecture with tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture with tutorials
Duration 1 semester
Frequency Regularly once a year
Language of English
instruction
Type of Exam Written test (in case of a small number of participants: oral tests)
Content This course equips students with concepts and tools that should be familiar to
anyone working with (large) data. Based on practical experiments and exam-
ples, frequently encountered pitfalls and problems are discussed alongside best
practices. We encounter basic statistical notions and problems of bias, testing
and experimental design. Foundational methods of machine learning and statis-
tical data analysis are employed to employ these ideas in practice. We will also
discuss best practices for scientific data presentation and documentation—how
to make expressive figures and tables and perform reproducible experiments—
and explore ethical and technical considerations in the context of fairness and
transparency.
Apart from mathmatical derivations, the exercises put a focus on practical
programming. In particular, they contain implementations of some content of
the lectures.
Objectives Students develop a sensitivity for common problems and misconceptions in
empirical work with data. They understand the mathematical, epistemologi-
cal, ethical, technical and social challenges surrounding the use of data, and
know best practices to address them. They also collect a concrete box of soft-
ware tools to collect, document, explore, visualize, and draw conclusions from
structured, large, small, corrupted and expensive data.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 2 3 W 90 g 100
Tutorial T o 2 3
Usability (modules) Diverse Topics in ML; General Computer Science; Expanded Perspectives
Requirement for only basic math and coding skills as provided by the BSc Computer Science.
participation
Lecturer Hennig
Literature Literature will be listed at the beginning of the semester.
Module handbook M.Sc. Machine Learning 18

Module Number: Module title Module

ML-4301 Numerical Algorithms of Machine Learning Lecture with tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture with tutorials
Duration 1 semester
Frequency irregularly
Language of English
instruction
Type of Exam Written test (in case of a small number of participants: oral tests)
Content The computational cost of machine learning is almost entirely caused by nu-
merical computations: Optimization for training and fitting of point estimates;
integration for marginalization and conditioning in probabilistic models; sim-
ulation, i.e. the solution of differential equations for predictions of the future,
and linear algebra as the base case of all of the above. These tasks are often
solved with “black-box” tools, but those who want to build highly performant,
scalable, professional solutions need to know how these tools worn and adapt
them to the specific task. This course introduces basic and advanced tools for
the aforementioned tasks. It develops a holistic view of computation in the
context of, and within the conceptual framework of machine learning, moving
from classic concepts to recent developments.
Apart from mathmatical derivations, the exercises put a focus on practical
programming. In particular, they contain implementations of some content of
the lectures.
Objectives Students develop both an intuitive and mathematical understanding of numer-
ical methods for optimization, integration, linear algebra, and the solution of
differential equation. They know how to adapt the tools to the challenges of
the task at hand, such as high dimensionality, stochasticity in computation,
numerical stability, non-convexity, efficient tuning of algorithmic parameters,
and uncertainty calibration for imprecise computation. Experience in the de-
sign and use of numerical tools is a highly sought-after skill in industry, and
distinguishes the expert engineer from the amateur user.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 2 3 W 90 g 100
Tutorial T o 2 3
Usability (modules) Diverse Topics in ML; General Computer Science; Expanded Perspectives
Requirement for Linear algebra is a core theme. Knowledge of probabilistic machine learning
participation is valuable for this course. Prior experience with numerical analysis is helpful
but not required. The practical parts use python and various recent python
libraries.
Lecturer Hennig
Literature Literature will be listed at the beginning of the semester.
Module handbook M.Sc. Machine Learning 19

Module Number: Module title Module

ML-4302 Statistical Learning Theory Lecture with Tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture with Tutorials
Duration 1 semester
Frequency Irregularly
Language of English
instruction
Type of Exam Written exam (in case of a small number of participants: oral exams)
Content Part 1: basic results in statistical learning theory:
• Statistical setup, estimation and approximation error, consistency
• Negative results: No free lunch theorem, slow rates of convergence
• Consistency of k nearest neighbor algorithms and partitioning algorithms

• Concentration inequalities
• Simple generalization bounds, for example with shattering coefficients
and VC dimension
• Advanced generalization bounds, for example using Rademacher com-
plexities, algorithmic stability, sample compression.
• Regularization and its consistency
Part 2: advanced results in statistical learning theory. This part of the lecture
changes, depending on the interests of the audience and the current state of
the art in the field and covers some of the recent results on learning theory. It
could cover topics like online learning, theory of unsupervised learning, theory
of deep learning, etc.
Objectives Students get to know the standard tools and approaches in statistical learning
theory. They understand positive and negative results in learning theory, in
particular what are the fundamental limitations of machine learning, and which
properties are important to make a machine learning algorithm work.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 2 3 W 90 g 100
Tutorial T o 2 3
Usability (modules) Diverse Topics in ML; General Computer Science; Expanded Perspectives
Requirement for Students need to know the contents of the basic math classes, in particular
participation linear algebra and probability theory.
Lecturer Ulrike von Luxburg
Literature The literature for this lecture will be provided at the beginning of the semester.
Module handbook M.Sc. Machine Learning 20

Module Number: Module title Module

ML-4303 Convex and Nonconvex Optimization Lecture with tutorials
ECTS 9
Work load
- Contact time Work load Class time Self-Study
- Self study 270 h 90 h / 6 CH 180 h
Lecture type Lecture with tutorials
Duration 1 semester
Frequency every two years
Language of English
instruction
Type of Exam Written test (in case of a small number of participants: oral tests)
Content Convex optimization problems arise quite naturally in many application areas
like signal processing, machine learning, image processing, communication and
networks and finance etc.
The course will give an introduction into convex analysis, the theory of convex
optimization such as duality theory, algorithms for solving convex optimiza-
tion problems such as interior point methods but also the basic methods in
general nonlinear unconstrained minimization, and recent first-order methods
in non-smooth convex optimization. We will also cover related non-convex
problems such as d.c. (difference of convex) programming, biconvex optimiza-
tion problems and hard combinatorial problems and their relaxations into con-
vex problems. While the emphasis is given on mathematical and algorithmic
foundations, several example applications together with their modeling as op-
timization problems will be discussed.
The course requires a good background in linear algebra and multivariate cal-
culus, but no prior knowledge in optimization is required.
Objectives Students learn the foundations of convex analysis and how to formulate and
transform optimization problems. After the lecture they know a variety of
methods for solving convex and non-convex optimization problems and have
guidelines which method to choose for which problem.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 4 6 W 90 g 100
Tutorial T o 2 3
Usability (modules) Diverse Topics in ML; General Computer Science; Expanded Perspectives
Requirement for Students need to know the contents of the basic math classes, in particular
participation linear algebra and multivariate calculus. No prior background in optimization
is required.
Lecturer Hein
Literature The lecture does not follow a specific book. The literature for this lecture will
be provided at the beginning of the semester.
Module handbook M.Sc. Machine Learning 21

Module Number: Module title Module

ML-4310 Data Mining and Probabilistic Reasoning Lecture with tutorials
ECTS 3
Work load
- Contact time Work load Class time Self-Study
- Self study 90 h 30 h / 2 CH 60 h
Lecture type Lecture with tutorials
Duration 1 semester
Frequency regularly in the winter term
Language of English
instruction
Type of Exam Written test (in case of a small number of participants: oral tests)
Content The lecture gives an introduction into the basics of probability theory, statis-
tics, information theory, data (pre-)processing and indexing techniques, graph
representations and link analysis, classification, clustering and topic models,
probabilistic inference in graphical models.
Objectives (1) The students acquire extensive knowledge in theory and application of
methods from the field of data science.
(2) The students acquire various data science techniques for conceptual think-
ing, problem formalization and problem solving.
(3) The students are introduced to challenging research questions from the field
of data science.

Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation
Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 1 2 W 90 g 100
Tutorial T o 1 1
Usability (modules) Diverse Topics in ML; General Computer Science; Expanded Perspectives
Requirement for
participation
Lecturer Gjergji Kasneci
Literature Will be supplied (book chapters and papers in English)
Module handbook M.Sc. Machine Learning 22

Module Number: Module title Module

ML-4320 Time Series Lecture with tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture with tutorials
Duration 1 semester
Frequency irregularly
Language of English
instruction
Type of Exam Written test (in case of a small number of participants: oral tests)
Content A time series is an extremely wide-spread type of empirical data: a (poten-
tially multivariate) set of observations that evolves over a univariate and thus
ordered index space—time. Examples include stock prices, inventory levels,
sports statistics, sensor readings in scientific equipment, cars and machinery,
and many more. Time series often require real-time processing, and can poten-
tially be infinitely long. But their univariate domain also allows for a crucial
property of the model: Markovianity, the ability to locally store all aspects of
the model necessary for inference in a time-local memory of fixed and finite
size. This course introduces a range of models and algorithms for efficient and
flexible inference in time series. Starting from famous concepts from the ar-
eas of signal processing and control, we will move to recent and contemporary
models for structured, high-dimensional, non-linear and irregular time series.
Alongside data and models, efficient algorithms for approximate inference are
a core focus.
Apart from mathmatical derivations, the exercises put a focus on practical
programming. In particular, they contain implementations of some content of
the lectures.
Objectives Students develop an understanding for key algorithmic and modelling chal-
lenges in the analysis of, and practical inference with time-ordered processes
and data. They can implement and debug basic and advanced models for
such data, including for production-level, large-scale applications, and for ar-
eas demanding high quality predictions, such as scientific analysis. Apart from
mathmatical derivations, the exercises put a focus on practical programming.
In particular, they contain implementations of some content of the lectures.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 2 3 wo 90 g 100
Tutorial T o 2 3
Usability (modules) Diverse Topics in ML;
Requirement for Knowledge of the material provided in the course Probabilistic Machine Learn-
participation ing (ML-4202) is required.f
Lecturer Hennig, Tronarp
Literature Literature will be listed at the beginning of the semester.
Module handbook M.Sc. Machine Learning 23

Module Number: Module title Module

ML-4340 Self-Driving Cars Lecture with tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture with tutorials
Duration 1 semester
Frequency Regularly once a year
Language of English
instruction
Type of Exam Written exam (in case of a small number of participants: oral exam)
Content Within the last years, driverless cars have emerged as one of the major
workhorses in the field of artificial intelligence. Given the large number of
traffic fatalities, the limited mobility of elderly and handicapped people as
well as the increasing problem of traffic jams and congestion, self-driving cars
promise a solution to one of our societies most important problems: the future
of mobility. However, making a car drive on its own in largely unconstrained
environments requires a set of algorithmic skills that rival human cognition,
thus rendering the task very hard. This course will cover the most dominant
paradigms of self-driving cars: modular pipeline-based approaches as well as
deep-learning based end-to-end driving techniques. Topics include camera, li-
dar and radar-based perception, localization, navigation, path planning, vehicle
modeling/control, imitation learning and reinfocement learning. The tutorials
will deepen the acquired knowledge through the implementation of several deep
learning based approaches to perception and sensori-motor control in the con-
text of autonomous driving. Towards this goal, we will build upon existing
simulation environments and established deep learning frameworks.
Objectives Students develop an understanding of the capabilities and limitations of state-
of-the-art autonomous driving solutions. They gain a basic understanding of
the entire system comprising perception, learning and vehicle control. In ad-
dition, they are able to implement and train simple models for sensori-motor
control.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 2 3 wt/ot 90 g 100

Tutorial T o 2 3
Usability (modules) Foundations of ML;
Requirement for Basic math (linear algebra, probabilities) and coding (Python) skills. Experi-
participation ence with deep learning (e.g., course “Deep Learning”). Experience in PyTorch
is recommended.
Lecturer Andreas Geiger
Literature Related literature will be listed throughout the lecture.
Module handbook M.Sc. Machine Learning 24

Module Number: Module title Module

ML-4360 Computer Vision Lecture with tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture with tutorials
Duration 1 semester
Frequency Regularly once a year
Language of English
instruction
Type of Exam Written exam (in case of a small number of participants: oral exam)
Content The goal of computer vision is to compute geometric and semantic properties of
the three-dimensional world from digital images. Problems in this field include
reconstructing the 3D shape of an object, determining how things are moving
and recognizing objects or scenes. This course will provide an introduction to
computer vision, with topics including image formation, camera models, cam-
era calibration, feature detection and matching, motion estimation, geometry
reconstruction, object detection and tracking, and scene understanding. Ap-
plications include building 3D maps, creating virtual avatars, image search,
organizing photo collections, human computer interaction, video surveillance,
self-driving cars, robotics, virtual and augmented reality, simulation, medical
imaging, and mobile computer vision. Modern computer vision relies heavily
on machine learning in particular deep learning and graphical models. This
course therefore assumes prior knowledge of deep learning (e.g., deep learning
lecture) and introduces the basic concepts of graphical models and structured
prediction where needed.
Objectives Students gain an understanding of the theoretical and practical concepts of
computer vision including image formation, camera models, feature detection,
multiple view geometry, 3D reconstruction, motion estimation, object recogni-
tion, scene understanding and structured prediction using deep neural networks
and graphical models. After this course, students should be able to understand
and apply the basic concepts of computer vision in practice, develop and train
computer vision models, reproduce research results and conduct original re-
search in this area.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 2 3 wt/ot 90 g 100

Tutorial T o 2 3
Usability (modules) Foundations of ML;
Requirement for Basic math (linear algebra, probabilities) and coding (Python) skills. Experi-
participation ence with deep learning (e.g., course “Deep Learning”). Experience in PyTorch
is recommended.
Lecturer Andreas Geiger
Literature Related literature will be listed throughout the lecture.
Module handbook M.Sc. Machine Learning 25

Module Number: Module title Module

ML-4601 Introduction to Game Theory with Application Lecture
to Multi-Agent Systems
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture
Duration 1 semester
Frequency regularly in the winter semester
Language of English
instruction
Type of Exam Written Exam
Content This module is about game theory and mechanism design, with an emphasis on
applications in different domains. The students study the essential concepts in
game theory such as equilibrium, belief, best-response dynamics, and the like.
Besides, they learn about strategic- and extensive form games, achieving equi-
librium in repeated games, games with incomplete and imperfect information.
Also, they obtain knowledge regarding other topics such as the Nash bargain-
ing solution, auctions, and computational models of human decision-making.
In brief, the students obtain broad knowledge about different branches of game
theory such as competitive-, cooperative-, and behavioral game theory, in addi-
tion to studying detailed mathematical results, e.g., regarding the existence and
uniqueness of equilibrium in well-known scenarios. Besides theoretical founda-
tions, the students become familiar with the connection between game theory
and distributed control, and they gain experience in modeling and solving dif-
ferent applied problems using game theory.
Objectives After the lectures, the students have a broad and profound knowledge of es-
sential concepts of game theory. Therefore, they can identify the problems in
the applied domains that can be modeled based on game theory. The students
possess the ability to solve such problems by using the mathematical tools that
they have learned in this module. Besides, they have a high level of proficiency
in selecting, reading, analyzing, and criticizing scientific results, preparing tech-
nical presentations, holding talks, and participating in discussions. Finally, the
students are independent learners and can expand their knowledge to advanced
levels in various topics of game theory.
Usability (modules) Diverse Topics in ML; General Computer Science; Expanded Perspectives
Requirement for
participation
Lecturer Maghsudi
Module handbook M.Sc. Machine Learning 26

(still ML-4601)
Literature

• Mas-Colell and M.D. Whinston, and J.R. Green, Microeconomic Theory,

Oxford University Press, 1995
• Ozduglar, Game Theory with Engineering Application, MIT Open-
CourseWare, 2009

• Fudenberg and D. Levine, The Theory of Learning in Games, MIT Press,

1998
• Fudenberg and J. Tirole, Game Theory, MIT Press, 1991
• Vijay, Auction Theory, Harvard University Press, 2008
Module handbook M.Sc. Machine Learning 27

Module Number: Module title Module

ML-4350 Reinforcement Learning Lecture, Tutorial
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture, Tutorial
Duration 1 semester
Frequency irregularly
Language of English
instruction
Type of Exam Oral presentation and written project report
Content

• Introduction to Machine Learn- tion and control

ing • Value Function Approximation
• Supervised Learning and Opti- • Policy Gradient
mization
• Deep RL, control in continuous
• Intro Reinforcement Learning state-action domains
(RL) and Markov Decision Pro- • Optimal Control and Model-
cesses based RL
• Dynamic Programming, predic- • Advanced topics in RL

Objectives ) Students can phrase a problem in the reinforcement learning framework and
can select an appropriate algorithm for solving it.
(2) Students are able to implement a set of deep reinforcement learning algo-
rithms and analyse their behavior.
(3) Students can explain the challenges in reinforcement learning and assess
and characterize new reinforcement learning methods.
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for Recommended to attend basic Machine learning class before.
participation
Lecturer Martius
Literature Reinforcement learning by Sutton and Barto http://incompleteideas.net/
book/bookdraft2017nov5.pdf
Pattern Recognition and Machine Learning by C.M. Bishop, Chap. 3 and 5
Deep Learning by Goodfellow, Bengio and Courville https://www.
deeplearningbook.org
Module handbook M.Sc. Machine Learning 28

Module Number: Module title Module

ML-4410 Neural Data Analysis Lecture, Tutorial
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 120 h 60 h / 4 CH 60 h
Lecture type Lecture, Tutorial
Duration 1 semester
Frequency regularly in the summer term
Language of English
instruction
Type of Exam Written report and cumulative oral exam
Content In recent years experimental methods to record brain activity have been rev-
olutionized. As the complexity of the data acquired in neuroscience increases,
neural data analysis becomes ever more important: The complex multidimen-
sional signals recorded with e.g. multielectrode arrays or two-photon imaging
can no longer be interpreted by eye, but rigorous data analytic techniques are
needed.
In this course we will cover a selection of topics related to the analysis of
different kinds of neural data based on concepts of machine learning: time
series analysis, spike sorting, spike triggered average/covariance, dimensionality
reduction techniques and information theory. The focus will be on applying
state-of-the-art concepts in hands-on data analysis of real data sets.

Objectives 1) In this course students will acquire knowledge of basic and advanced tech-
niques necessary to analyze discrete (spike trains) and continuous (cellular volt-
age/calcium signals, LFP, EEG) neural signals. (2) Students will implement
important techniques (Filtering, MoG, STA, etc) and evaluate them on artifi-
cial and real data. (3) Students will learn how to work with real neural data
and cope with the challenges this brings about.
Usability (modules) Diverse Topics in ML; General Computer Science; Expanded Perspectives
Requirement for Some knowledge of basic neuroscience is helpful, but not a must.
participation
Lecturer Berens
Literature Emery N Brown, Robert E Kass, und Partha P Mitra, „Multiple neural spike
train data analysis: state-of-the-art and future challenges“, Nat Neurosci 7, Nr.
5 (Mai 2004): 456-461.
Robert E. Kass, Valérie Ventura, und Emery N. Brown, „Statistical Issues in
the Analysis of Neuronal Data“, Journal of Neurophysiology 94, Nr. 1 (Juli 1,
2005): 8 -25.
Dayan and Abbott: Theoretical Neuroscience. MIT Press.
Rieke, Warland, Ruyter van Stevenik and Bialek: Spikes – Exploring the neural
code. MIT Press.
Module handbook M.Sc. Machine Learning 29

Module Number: Module title Module

INFO-4492 Special Topics in Learning Theory Lecture, Tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture, Tutorials
Duration 1 semester
Frequency irregularly
Language of English or German, depending on the participants
instruction
Type of Exam Written exam (in case of a small number of participants: oral exams)
Content In this module we discuss advanced results and approaches in learning theory
and current research results in the area of machine learning in general.
Objectives Students get to know about advanced results in learning theory. They can
judge whether an algorithm is well designed, both from an algorithmic and
statistical point of view. They understand about the fundamental limitations
of machine learning. They can reflect current research questions. After this
module they are well-prepared to write a master thesis in the area of learning
theory.

Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation
Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 2 3 K 90 b 100
Tutorial T o 2 3
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for Solid knowledge in maths (linear algebra, probability theory); Basic knowledge
participation in machine learning
Lecturer von Luxburg
Literature will be announced in the lecture
Module handbook M.Sc. Machine Learning 30

Module Number: Module title Module

ML-4420 Efficient Machine Learning in Hardware Lecture
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 32 h / 4 CH 208 h
Lecture type Lecture
Duration 1 semester
Frequency regularly in the summer, every two years
Language of English
instruction
Type of Exam Oral exam
Content The recent breakthroughs in using deep neural networks for a large variety of
machine learning applications have been strongly influenced by the availability
of high performance computing platforms. In contrast to its biological origin,
however, high performance of artificial neural networks critically relies on much
higher energy demands. While the average energy consumption of the entire
human brain is comparable to that of a laptop computer (i.e. 20W), artificial
intelligence often resorts to large HPCs with several orders of magnitude higher
energy demand. This lecture will discuss this problem and show solutions on
how to build energy and resource efficient architectures for machine learning
in hardware. In this context, the following topics will be addressed:
• Hardware architectures for machine learning: GPU, FPGA, SIMD ar-
chitectures, domain-specific architectures, custom accelerators, in/near
memory computing, training vs. inference architectures
• Energy-efficient machine learning
• Optimized mapping of deep neural networks to hardware and pipelining
techniques
• Word length optimization (binary, ternary, integer, floating point)
• Scalable application specific architectures
• New switching devices to implement neural networks (Memristors, PCM)
• Neuromorphic computing

Objectives The students gain in-depth knowledge about the challenges associated with
energy-efficient machine learning hardware and respective state-of-the-art solu-
tions. They can compare different hardware architectures regarding the trade-
off between energy consumption, complexity, computational speed and the
specificity of their applicability. The students learn what kinds of hardware
architectures are used for machine learning, understand the reasons why a par-
ticular architecture is suitable for a particular application, and can efficiently
implement machine learning algorithms in hardware.
Usability (modules) Diverse Topics in ML; General Computer Science; Expanded Perspectives
Requirement for Knowledge about foundations in machine learning
participation
Lecturer Bringmann
Literature Will be announced in the first lecture
Module handbook M.Sc. Machine Learning 31

Seminars

Module Number: Module title Module

ML-4501 Machine Learning Seminar Seminar
ECTS 3
Work load
- Contact time Work load Class time Self-Study
- Self study 90 h 30 h / 2 CH 60 h
Lecture type Seminar
Duration 1 semester
Frequency regularly in the winter/summer
Language of English
instruction
Type of Exam Oral presentation and written report
Content In this module we discuss advanced results and approaches in machine learning
theory and application and current research results in the area of machine
learning in general.
Objectives Students get to know about advanced results in machine learning theory and
applications. They can judge for example whether an algorithm is well de-
signed, both from an algorithmic and statistical point of view. They under-
stand about the fundamental limitations of machine learning. They can reflect
current research questions. Students will be able to acquire knowledge about
current findings through comprehensive literature search. They will know the
importance of current topics in the area of machine learning, and will be aware
that there are still many open questions. Students will not only have improved
their study and reading skills, but will also have enhanced their capability of
working independently. The teaching method in this seminar aims at boosting
the students’ confidence (oral presentation), and at enhancing their communi-
cation skills and enabling them to accept criticism (discussion session following
their presentation. After this module they are well-prepared to write a master
thesis in the area of machine learning.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)

Requirement for
Evaluation

Credit Points /
Status

Grade
CH

CP

Seminar S o 2 3 wo 30 g 100
Usability (modules) Diverse Topics in ML; General Computer Science; Expanded Perspectives
Requirement for
participation
Lecturer All lecturers in the computer science department
Literature Will be handed out in the course
Module handbook M.Sc. Machine Learning 32

Module Number: Module title Module

ML-4503 Explainable Machine Learning Seminar
ECTS 3
Work load
- Contact time Work load Class time Self-Study
- Self study 90 h 30 h / 2 CH 60 h
Lecture type Seminar
Duration 1 semester
Frequency regularly in the winter semester
Language of English
instruction
Type of Exam Oral Presentation (about 30 minutes) and written elaboration (approx. 10
pages), leading the discussion once
Content

• In this seminar, we will discuss research papers related to explainable

machine learning focusing on generating visual and textual explanations
for classification decision of machine learning models.
• From a methodological perspective, we will discuss about popular per-
ceptual modules of machine learning models, integrated attention mech-
anisms as well as memory based natural language processing methods
tailored towards explanation generation.

• General knowledge on Statistical • General knowledge on Computer

Machine Learning Vision
• General knowledge on Deep • General knowledge on Natural
Learning Language Processing is a plus

Objectives Students are able to read and reflect upon current research papers in this
research area. They can critically assess the contributions of such a paper.
They can present current research results to other students and researchers
and can lead research discussions. The form of learning used in the seminar is
intended to help the students to develop self-confidence (presentation) and the
ability to criticise and communicate (in the subsequent discussion).
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)

Requirement for
Evaluation

Credit Points /
Status

Grade
CH

CP

Seminar S o 2 3 wo 30 g 100
Usability (modules) Diverse Topics in ML; General Computer Science; Expanded Perspectives
Requirement for
participation
Lecturer Akata
Literature Will be announced in the first meeting
Module handbook M.Sc. Machine Learning 33

Module Number: Module title Module

INFO-4493 Learning Theory Seminar
ECTS 3
Work load
- Contact time Work load Class time Self-Study
- Self study 90 h 30 h / 2 CH 60 h
Lecture type Seminar
Duration one semester
Frequency irregularly
Language of English or German, depending on the participants
instruction
Type of Exam Oral presentation, written report.
Content In this seminar we discuss current research papers in the area of machine learn-
ing theory, in the form of student’s presentations and guided discussions.
Objectives Students are able to read and reflect upon current research papers in the area
of learning theory. They can critically assess the contributions of such a paper.
They can present current research results to other students and researchers and
can lead research discussions. They can summarize and evaluate the results of
a paper in form of a written research report.

Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Requirement for

Evaluation
Credit Points /
Status

Grade
CH

CP

Seminar S o 2 3 R 45 b 100
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for Basic knowledge in machine learning.
participation
Lecturer von Luxburg
Literature will be announced in the lecture
Module handbook M.Sc. Machine Learning 34

Module Number: Module title Module

ML-4502 Machine learning methods for scientific discovery Seminar
ECTS 3
Work load
- Contact time Work load Class time Self-Study
- Self study 90 h 30 h / 2 CH 60 h
Lecture type Seminar
Duration 1 semester
Frequency irregularly
Language of English
instruction
Type of Exam Oral presentation, written report
Content In this seminar, we will discuss current and classical research papers which
describe machine learning methods for applications in the natural sciences.
From a methodological perspective, a particular focus will be on ‘simulation-
based inference approaches’, as these provide a bridge between data-driven
machine learning methods, and theory-driven scientific modelling, as well as
on latent-variable models for inferring dynamical systems from data.
Objectives Students are able to read and reflect upon current research papers in this
research area. They can critically assess the contributions of such a paper.
They can present current research results to other students and researchers and
can lead research discussions. They can summarize and evaluate the results of
a paper in form of a written research report.

Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Requirement for
Credit Points / Evaluation
Status

Grade
CH

CP

Seminar S o 2 3 wo 30 g 100
Usability (modules) Diverse Topics in ML; General Computer Science; Expanded Perspectives
Requirement for Basic knowledge probabilistic machine learning
participation
Lecturer Macke
Literature Will be announced in the first meeting
Module handbook M.Sc. Machine Learning 35

Practical Courses and Research Project

Module Number: Module title Module

ML-4510 Practical Machine Learning Practical Course
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Practical Course
Duration 1 semester
Frequency every semester
Language of English
instruction
Type of Exam Oral presentation, written report, lab journal
Content The practical course consists of finishing assigned tasks in small teams, au-
tonomously or under supervision. Study and exam performance are usually
evaluated based on active participation, a presentation of results and in writ-
ten reports.
Objectives Students will gain practical experience in designing and programming methods
/ software /tools for ML. They will be able to use libraries and frameworks, and
will acquire knowledge or extend their knowledge of various programming lan-
guages. By working together in groups, students obtain teamwork and collab-
oration skills, and they will learn about project organization and presentation
techniques. Students will know about the strengths and weaknesses and about
the limitations of various methods for evaluating complex and high-dimensional
data, and will be able to describe and evaluate these methods.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Requirement for
Evaluation

Credit Points /
Status

Grade
CH

CP

Practical P o 4 6 wo g 100
Usability (modules) Diverse Topics in ML; General Computer Science; Expanded Perspectives
Requirement for
participation
Lecturer All lecturers in the programme
Literature -
Module handbook M.Sc. Machine Learning 36

Module Number: Module title Module

ML-4998 Research Project Machine Learning Independent research
project
ECTS 9
Work load
- Contact time Work load Class time Self-Study
- Self study 270 h 30 h / 2 CH 240 h
Lecture type Independent research project
Duration 1 semester
Frequency each semester
Language of English
instruction
Type of Exam Essay
Content The research project serves to deepen theoretical and practical knowledge in a
specific field of machine learning. Students are working on a research project
with the main focus of the research group.
Objectives The students
• get an insight into scientific work,

• learn how to independently pursue a research question,

• learn independently to identify and compile scientific literature for the
question to be worked on,
• are able to work in a team in an international scientific environment,

• deepen their problem-solving skills,

• can give a scientific lecture
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Requirement for
Evaluation

Credit Points /
Status

Grade
CH

CP

Research Project R o 2 9 tp g 100

Usability (modules) Diverse Topics in ML;
Requirement for Excellent academic grades in Master Machine Learning. There are only a few
participation research projects that are offered semester by semester. A written application,
including letter of motivation, CV and Transcript of Records should be sent to
the research group leader of the offered research project.
Lecturer All professors in Machine learning
Literature Scientific literature/publications relevant to the research topic to be addressed
Study Area: General Computer
Science

Module Number: Module title Module

INFO-4315 Advanced Topics in Embedded Systems Lecture
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 30 h / 2 CH 150 h
Lecture type Lecture
Duration 1 semester
Frequency regularly in the summer term (block course)
Language of English
instruction
Type of Exam Oral exam (written exam in case of a large number of participants)
Content This lecture discusses current topics and trends in embedded system research
with special focus on design, analysis and verification of embedded systems and
Systems-on-Chip (SoCs). The lecture starts with an introduction into embed-
ded systems architectures and electronic system level design. Then, the latest
developments in analysis of non-functional properties like timing, power dis-
sipation, and energy consumption are discussed. The lectures on verification
addresses cyber-physical systems, safety verification, and robustness optimiza-
tion of machine-learning based embedded systems. The lecture finally covers
advanced hardware architectures for low-power implementation of deep learn-
ing approaches in hardware. Between the lectures, practical exercises in form
of programming assignments will take place. The lecturers will present the
relevant basics as well as recent research results in each topic.
Objectives Participants will acquire in-depth knowledge to different aspects in embedded
systems as well as the necessary skills to design, analyse, and verify embed-
ded systems under safety constraints. They will gain hands-on experience in
embedded system design in order to avoid common pitfalls. The students will
get a deeper practical understanding by working on topic-specific programming
assignments.

37
Module handbook M.Sc. Machine Learning 38

(still INFO-4315)

Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation
Requirement for
Credit Points /

Status
Grade

CH

CP
Lecture L o 2 4 MP 30 b 100
Practical P o 2 2
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for Prerequisites are the lectures “Entwurf und Synthese Eingebetteter Systeme”
participation or “Modellierung und Analyse Eingebetteter Systeme”
Lecturer Bringmann
Literature Will be announced during the first lecture.
Module handbook M.Sc. Machine Learning 39

Module Number: Module title Module

INFO-4194 Behavior and Learning Lecture, Tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 32 h / 4 CH 208 h
Lecture type Lecture, Tutorials
Duration 1 semester
Frequency irregularly
Language of English
instruction
Type of Exam Written (oral exam if number of participants allows)
Content This lecture builds on the available knowledge how animals and humans plan,
decide on, and control their behavior and how they progressively optimize and
adapt their behavior over time. Accordingly, algorithms are introduced for
behavioral decision making, control, optimization, and adaptation. In par-
ticular, the lecture introduces spatial representations for behavioral control,
forward-inverse control models, including the learning of such representations
and models. Also the encoding and the learning of motor control primitives
and motor complexes is considered. Last but not least, self-motivated artifi-
cial systems are considered that strive to maintain internal homeostasis and to
maximize information gain.
Objectives Students know how intelligent behavior can be generated and learned in artifi-
cial systems. They can apply reinforcement learning (RL), including hierarchi-
cal RL, factored RL, and actor-critic approaches to the appropriate problems.
Moreover, they are aware of the contrast between model-free and model-based
RL approaches. They know about dynamic motion primitives and know how
to optimize them. Moreover, they know about Gaussian Mixture Models, in-
cluding how to learn and optimize them. They can implement information-
gain driven and self-motivated behavior and are aware of the exploration-
exploitation dilemma. Moreover, they are aware of model-predictive control, of
options to learn suitable model-predictive structures, and of options to suitably
abstract such structures. Finally, they know how sensorimotor-grounded spa-
tiotemporal representations can be learned, stored as episodic memory units,
and can be abstracted into cognitive maps, enabling model-based RL.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 2 3 wt 90 g 100
Tutorial T o 2 3 0
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for Introductory course about machine learning, artificial neural networks,
participation robotics, or artificial intelligence is required.
Lecturer Butz
Literature Will be supplied (book chapters and papers in English)
Module handbook M.Sc. Machine Learning 40

Module Number: Module title Module

INFO-4210 Recurrent and Generative Artificial Neural Net- Lecture, Tutorials
works
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 32 h / 4 CH 208 h
Lecture type Lecture, Tutorials
Duration 1 semester
Frequency irregularly
Language of English
instruction
Type of Exam Written (oral exam if number of participants allows)
Content Advanced ANN topics. First, revisiting backpropagation and backpropaga-
tion through time; then: Advanced Recurrent Neural Networks (LSTM, GRU);
Very Deep Learning and Generative Adversarial Networks; Spatial and Tempo-
ral Convolution; Reservoir Computing; Neuroevolution; Attention and Routing
Networks; Autoencoders and Restricted Boltzmann Machines; Gain Fields and
Switching Networks; Latent Space Visualization techniques; Generative Infer-
ence
Objectives Students know about and how to apply generative and typically recurrent ar-
tificial neural networks in various domains including data classification, image
recognition, language processing, spatially-invariant recognition, spatial trans-
formations, and spatial mappings. They can apply complex, generative artifi-
cial neural networks from scratch as well as with available tools. They know
how to optimize weights and network structures by means of gradient descent
as well as by alternative methods. They can use complex recurrent network
structures to selectively process aspects of the data. They know how to apply
generative networks as model-predictive neural controllers and as well as long-
range temporal predictors. They can combine retrospective latent state and
motor inference techniques with prospective motor control.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 2 3 wt 90 g 100
Tutorial T o 2 3 0
Usability (modules) Diverse Topics in ML; General Computer Science; Expanded Perspectives
Requirement for Knowledge about machine learning, artificial neural networks, deep learning,
participation or artificial intelligence is required.
Lecturer Butz
Literature Will be supplied (book chapters and paper in English)
Module handbook M.Sc. Machine Learning 41

Module Number: Module title Module

INFO-4212 Artificial Neural Networks Practical Course
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Practical Course
Duration 1 semester
Frequency irregularly
Language of English
instruction
Type of Exam Final Project Presentation and Report
Content Programming enhanced functionalities in ANN Software, evaluating perfor-
mance, analyzing the system.
Objectives Know how to work with, implement, and enhance complex artificial neural
networks..

Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Requirement for

Evaluation
Credit Points /
Status

Grade
CH

CP

Practical P o 2 3 tp g 100
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for Solid Knowledge in Programming. Knowledge about artificial neural networks
participation or machine learning.
Lecturer Butz
Literature none
Module handbook M.Sc. Machine Learning 42

Module Number: Module title Module

INFO-4213 Advanced Artificial Neural Networks Project Practical Course
ECTS 3
Work load
- Contact time Work load Class time Self-Study
- Self study 90 h 30 h / 2 CH 60 h
Lecture type Practical Course
Duration 1 semester
Frequency irregularly
Language of English
instruction
Type of Exam Final Project Presentation and Report
Content Working with ANN Software, evaluating performance, & analyzing the system.
Objectives Know how to evaluate, program, and analyze artificial neural networks.

Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Requirement for

Evaluation
Credit Points /
Status

Grade
CH

CP

Practical P o 2 3 tp g 100
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for Solid Knowledge in Programming. Knowledge about artificial neural networks
participation or machine learning.
Lecturer Butz
Literature none
Module handbook M.Sc. Machine Learning 43

Module Number: Module title Module

INFO-4214 Cognitive Modeling Lecture, Tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 32 h / 4 CH 208 h
Lecture type Lecture, Tutorials
Duration 1 semester
Frequency irregularly
Language of English
instruction
Type of Exam Written (oral exam if number of participants allows)
Content Cognitive models covering learning, action and perception are presented and
discussed, including descriptive, qualitative, quantitative and neural models. In
addition, parameter optimization as well as techniques to compare models and
to interpret and evaluate model parameters are introduced. All techniques are
shown in the context of concrete models of cognitive processes. Moreover, the
necessary statistical methods are introduced in a practical, application-oriented
manner.
Objectives Students know the most important principles and techniques of cognitive mod-
eling. They know how to model cognitive processes, mechanisms, and learning
at different levels of complexity. They can apply various cognitive models and
modeling approaches in a goal-directed manner. Moreover, they can evaluate,
compare, and contrast different modeling approaches as well as modeling re-
sults. They are able to judge whether a model is falsifiable and they know how
to validate and interpret cognitive models. Finally, they can use statistical
methods to quantitatively compare different cognitive models.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 2 3 wt 90 g 100
Tutorial T o 2 3 0
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for Introductory course knowledge about machine learning, artificial neural net-
participation works, robotics, cognitive architectures, or artificial intelligence is required.
Lecturer Butz
Literature Book: S. Lewandowsky & S. Farrell (2011). Computational Modeling in Cog-
nition. Additional papers and book chapters will be supplied.
Module handbook M.Sc. Machine Learning 44

Module Number: Module title Module

INFO-4211 Avatars in Virtual Realities Practical Course
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Practical Course
Duration 1 semester
Frequency Irregularly
Language of English
instruction
Type of Exam Final Project Presentation and Report
Content In this project-oriented practical course, students learn how to design realis-
tic, interesting, behaving avatars in virtual realities. Typically the focus lies
in developing user interfaces, and new options for interacting with the VR
and acting upon objects or other entities within the VR. Alternatively, exper-
imental setups will be programmed and optimized in order to run real-world
psychological and evaluative experiments in which users control avatars in VR.
Objectives Students know how to work with virtual realities (VRs) and how to develop
animated, autonomous avatars in these environments. They are able to create
and use suitable interfaces to enable users to effectively interact with VRs and
control avatars within.

Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Requirement for

Evaluation
Credit Points /
Status

Grade
CH

CP

Practical P o 4 6 tp g 100
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for Solid Knowledge in Programming. General knowledge about simulation soft-
participation ware.
Lecturer Butz
Literature none
Module handbook M.Sc. Machine Learning 45

Module Number: Module title Module

INFO-4250 Information Processing for Perception and Action Seminar
ECTS 3
Work load
- Contact time Work load Class time Self-Study
- Self study 90 h 30 h / 2 CH 60 h
Lecture type Seminar
Duration 1 semester
Frequency irregularly
Language of Deutsch, English
instruction
Type of Exam Wird zu Beginn des Semesters bekanntgegeben / Will be announced at begin-
ning of semester
Content Humans as well as complex technical systems process sensory information to
interact with the environment. These actions have consequences which (again)
create sensory events that can be processed and used to improve the interaction
with the environment. We will discuss advanced topics of this full ’perception–
action’ loop; in humans as well as in technical systems. A special focus will be
on the experimental literature from the Cognitive– and Neurosciences and on
advanced statistical methods.
Objectives Students will know current views on biological information processing and on
the interaction of humans with technical systems. They will also learn and
understand advanced statistical and empirical methods that were used to gen-
erate this knowledge. This expertise will help them to apply their knowledge in
interdisciplinary working environements, whenever empirical studies on human
performance and actions are required.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Requirement for
Evaluation

Credit Points /
Status

Grade
CH

CP

Seminar S f 2 3 tp 45 g 100
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for No formal requirements, but students should have a good background in statis-
participation tics and should have attended introductory/mid–level courses in Cognitive Sci-
ence/Neuroscience.
Lecturer Franz
Literature Wird zu Beginn des Semesters bekanntgegeben / Will be announced at begin-
ning of semester
Module handbook M.Sc. Machine Learning 46

Module Number: Module title Module

INFO-4152 Advanced Statistics Lecture, Tutorials
ECTS 3
Work load
- Contact time Work load Class time Self-Study
- Self study 90 h 2 h / 2 CH 88 h
Lecture type Lecture, Tutorials
Duration 1 semester
Frequency irregularly
Language of English
instruction
Type of Exam Pass/fail depending on performance in homework (every 4. session is a tutorial
for which we expect participants to have prepared and handed in homework;
typically some implementation in R/SPSS; for each session we expect partici-
pants to have read the relevant literature).
Content Advances in neuroscientific methodology give rise to the accumulation of huge
amounts of data. Analysing these data poses new problems that are typically
not covered by the classical introductory statistics courses and also increase the
need to master classic statistical topics as, for example, statistical power and
required sample sizes, problems of multiple testing, correlational structure of
repeated measures, etc. In short, solid statistical knowledge beyond standard
tests and ANOVAs are very important for anyone working in the neurosciences
today.
Moreover, in recent years, alternative approaches to data analysis have re-
ceived increasing attention because they can solve specific problems and incon-
sistencies of classical statistics. E.g. Bayesian approaches makes use of our
previous knowledge about the data, and non-parametric permutation statis-
tics/Bootstrap have the advantage of being relatively free of assumptions about
the underlying distribution of the data. This course will present these statisti-
cal methods in a way that focuses on understanding the guiding principals as
well as the practical applications of these methods in real neuroscientific data.
Further details: http://www.ecogsci.cs.uni-tuebingen.de/teach.php
Objectives Being able to understand and apply somewhat advanced statistical methods to
empirical research questions in the life-sciences/neuroscience.
Usability (modules)
Requirement for Basic/intermediate knowledge of classic statistics. You should feel comfort-
participation able with basic statistical topics as between-groups ANOVA, t-tests, regression
analysis, basics of repeated-measures ANOVA, and the rationale/mathematics
behind these procedures. You should also feel comfortable (or be willing to
learn rapidly) with implementing these basic methods either in the program-
ming language R or in the SPSS macro language (‘syntax mode’).
Lecturer Franz and Gaiss (Medical Faculty)
Literature Literature will be announced during the course.
Module handbook M.Sc. Machine Learning 47

Module Number: Module title Module

INFO-4149 Selected Topics in Database Systems Lecture, Tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture, Tutorials
Duration one semester
Frequency irregularly
Language of English (or German, depends on participants)
instruction
Type of Exam Written exam (oral exam for small number of participants), exercise points can
be included in the exam as bonus points.
Content Changing in-depth topics from the various areas of the Research field database
systems. Use of database systems for the realisation of demanding applications
(Advanced SQL).
Objectives The students have knowledge of research methodology in the field of database
systems. The focus is mainly on the use of SQL as database language, their
efficient translation, as well as their use for Implementation of very complex
applications. The participants are familiar with the preparation of scientific
papers, particularly in sub-areas of the research field of database systems Stu-
dents can focus specifically on Master’s theses and research projects.

Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation
Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 2 4 W 90 g 100
Tutorial T o 2 2
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for INF3131 Introduction to Relational Database Systems (DB1)
participation
Lecturer Grust
Literature Classical and current research literature on the subject area.
Module handbook M.Sc. Machine Learning 48

Module Number: Module title Module

INFO-4381 Advanced Topics in Human-Computer Interac- Seminar
tion
ECTS 3
Work load
- Contact time Work load Class time Self-Study
- Self study 90 h 30 h / 2 CH 60 h
Lecture type Seminar
Duration one semester
Frequency irregularly
Language of English
instruction
Type of Exam Oral presentation of at least 30 minutes and written report (essay at least 8
pages)
Content This seminar covers current and varying topics from research and application
in the field of (multimodal) human-machine interaction.
Objectives Students will read and reflect upon current research in the area of human-
computer interaction. They can present current research results to other stu-
dents and researchers as well as lead research discussions. They can summarize
and evaluate the results of a paper in the form of a written research report.

Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Requirement for

Evaluation
Credit Points /
Status

Grade
CH

CP

Seminar S o 2 3 tp 30 g 100
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for none
participation
Lecturer Kasneci
Literature none
Module handbook M.Sc. Machine Learning 49

Module Number: Module title Module

INFO-4412 Algorithms and Complexity Lecture, Tutorial
ECTS 9
Work load
- Contact time Work load Class time Self-Study
- Self study 270 h 90 h / 6 CH 180 h
Lecture type Lecture, Tutorial
Duration 1 semester
Frequency regularly in the winter, every year
Language of English
instruction
Type of Exam Written Exam (Oral Exam at small number of participants), grades in the
tutorial might be included to the final grade as bonus
Content Topics amongst others are:

• Matching
• MinCostFlow
• Approximation Schemes
• Network Analysis
• Clustering
• Algorithmic Geometry
• Discussions about complexity, e.g. lower bounds

Objectives Students gain in-depth knowledge about algorithmic techniques in different

fields of problems. This includes the application of sophisticated graph algo-
rithms, the proficiency in strategies for network analysis as well as the ability
to apply and develop approximation methods. Regarding the field of complex-
ity, the students can judge the difficulty level of problems and also prove their
judgements by techniques learned in this course.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 4 6 W 90 g 100
Tutorial T o 2 3
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for
participation
Lecturer Kaufmann
Literature Raghavan, Magnati, Orlin: Randomized Algorithms
Mehlhorn, Näher: LEDA - A platform for combinatorial and geometric com-
putation
Papadimitriou, Steiglitz: Combinatorial optimization : algorithms and com-
plexity
Module handbook M.Sc. Machine Learning 50

Module Number: Module title Module

INFO-4241 Programming Languages II Lecture, Tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 32 h / 4 CH 208 h
Lecture type Lecture, Tutorials
Duration 1 semester
Frequency about every two years
Language of English or German dependent on participants)
instruction
Type of Exam Written or oral examination. Participation in exercises is required for exam
participation.
Content This lecture is about the semantics and type systems of modern programming
languages. We discuss the foundations of programming languages using formal
semantics (such as small-step operational semantics), formal type systems and
their properties, and different variants of typed lambda calculi that constitute
the foundation for modern type systems.
Objectives Students will be able to discuss and analyze modern programming languages in
terms of the properties of their theoretical foundations. They will understand
the design space and tradeoffs of type systems for these languages.
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for Programming Languages I is helpful, but not required.
participation
Lecturer Ostermann
Literature Benjamin C. Pierce. Types and Programming Languages. MIT Press, 2003.
Module handbook M.Sc. Machine Learning 51

Module Number: Module title Module

INFO-4246 Programming with Dependent Types Practical Course
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 120 h 60 h / 4 CH 60 h
Lecture type Practical Course
Duration 1 semester
Frequency irregularly
Language of English or German, depends on participants
instruction
Type of Exam Project 50 %, Presentation and Documentation 50 %
Content Dependent types are types that can depend on values: arrays of length 25, 20-
by-20 matrices, or integers larger than -3. Agda and Idris are two dependently
typed languages. Dependent types are good for many things -— from elim-
inating ArrayIndexOutOfBoundsException to mechanically verified programs
and proving mathematical theorems -— yet the idea itself originates from the
foundational crisis of mathematics at the turn of the 19th century. Dependent
types carry their own coding patterns and caveats. In this seminar, we learn to
program effectively with dependent types: How to make hard things possible,
how to not make simple things hard, and a bit of how things work under the
hood.
Objectives The students can use a dependently-typed language such as Agda or Coq and
use dependent types to express and prove non-trivial program properties.
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for Participation in Programming Languages I, II or III is helpful but not required.
participation
Lecturer Ostermann
Literature will be announced at beginning of course
Module handbook M.Sc. Machine Learning 52

Module Number: Module title Module

INFO-4248 Interactive Theorem Proving Lecture, Tutorials
ECTS 9
Work load
- Contact time Work load Class time Self-Study
- Self study 270 h 34 h / 6 CH 296 h
Lecture type Lecture, Tutorials
Duration 1 semester
Frequency about every two years
Language of Englisch, if all participants agree, else German
instruction
Type of Exam Written or oral examination. Participation in exercises is required for exam
participation.
Content This course is an introduction to interactive theorem programming and ad-
vanced functional programming, mostly using the Coq proof assistant.
This course is for students interested in:
1. The foundational theories of mathematics, most notably type theory and
logic
2. Practical interactive theorem proving in a state-of-the-art proof assistant

3. Advanced functional programming languages and their relation to con-

structive mathematics via the “Curry-Howard Isomorphism”
4. Program verification and “certified programming”
5. Programming Language Semantics

Objectives Students will be able to write programs and prove theorems in the Coq proof
assistant. Students understand the theoretical underpinnings of interactive
theorem provers and get basic insights into the semantics and formal properties
of programming languages.
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for A background in functional programming is helpful. Experience with mathe-
participation matical proofs is helpful.
Lecturer Ostermann
Literature Volume 1 and 2 of the “Software Foundations” series available at https://
softwarefoundations.cis.upenn.edu/.
A. Chlipala, Certified Programming with Dependent Types, MIT Press
Module handbook M.Sc. Machine Learning 53

Module Number: Module title Module

BIO-4242 Advanced Java in Bioinformatics Lecture and tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture and tutorials
Duration 1 semester
Frequency every two years
Language of English
instruction
Type of Exam Programming project
Content In this course, we study the latest features of Java to address challenging pro-
gramming problems in bioinformatics. Topics include JavaFx, two- and third-
dimensional graphics, properties and bindings, animation, concurrent program-
ming and webprogramming. We will build a full-featured, interactive bioinfor-
matics program.

Objectives The students are able to design and implement a fully featured bioinformatics
program. They are able to analyze a computational problem and to develop
an appropriate solution. They are aware of both the possibilities and the
limitations of the application of Java to solve computational tasks. They are
able to analyse problems on a scientific level and summarise them in writing.
In particular, a high degree of intrinsic motivation and personal responsibility
is encouraged. Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Requirement for
Evaluation

Credit Points /
Status

Grade
CH

CP

Lecture L o 4 6 W 90 g 100
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for BIOINF4110 Sequence Bioinformatics
participation
Lecturer Huson
Literature Programming and bioinformatics literature
Module handbook M.Sc. Machine Learning 54

Module Number: Module title Module

BIO-4311 Microbiome analysis Lecture, Tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture, Tutorials
Duration 1 semester
Frequency every two years
Language of English
instruction
Type of Exam Written or oral exam
Content This course provides an in-depth introduction to microbiome analysis. Topics
include: Sequencing technologies. Community profiling using the SSU rRNA
gene. Community profiling using shotgun sequencing. Alignment-free and
alignment-based taxonomic profiling. Functional analysis and profiling. Sam-
ple comparison and time-series analysis.
Objectives The students are familiar with recent bioinformatics findings on microbiome
analysis. They can formulate the challenges of microbiome analysis for bioin-
formatics. They know algorithms for taxonomic and functional analysis of mi-
crobiome sequencing data, statistical methods for comparison and methods for
community profiling using 16S sequences. Students can analyse microbiome
sequencing data and perform profiling and comparison. They are aware of
both the possibilities and the limitations of different methods in this subfield
of bioinformatics. They are able to analyse problems on a scientific level and
summarise them in writing. In particular, a high degree of intrinsic motivation
and personal responsibility is encouraged.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 2 4 W 90 g 100
Tutorial T o 2 2
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for BIO-4110 Sequence Bioinformatics
participation
Lecturer Huson
Literature Lecture notes and scientific publications
Module handbook M.Sc. Machine Learning 55

Module Number: Module title Module

BIO4364 Visualization of Biological Data Lecture, Tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture, Tutorials
Duration 1 semester
Frequency Regularly in the winter semester
Language of English
instruction
Type of Exam Oral Exam (or written exam if number of participants is large)
Content As biological datasets increase in size and complexity, we are moving more
and more from an hypothesis-driven research paradigm to a data-driven one.
As a result, the visual exploration of that data has become even more crucial
than in the past. The aim of this lecture is to familiarize the participants
with modern methodologies of Information Visualization and Visual Analytics.
Information Visualization is concerned with methods for the visualization of
abstract data that has no inherent spatial structure (the visualization of spatial
data is covered in INF3145 - Scientific Visualization). The lecture imparts
how to apply these methods to biological data using practical examples and
provides hands-on training during the tutorials. Questions such as ‘what is data
visualization’, ‘what is visual analytics’, and ‘how can we visualise (biological)
data to gain insight in them, so that hypotheses can be generated or explored
and further targeted analyses can be defined’ are discussed. No prior knowledge
of biology is required, that is, the lecture is also suitable for students from other
fields such as computer science or media/medical informatics.
Objectives Students understand the visual analysis process. They know basic methods of
information visualization and the ‘do’s’ and ‘don’ts’ of visualization. The know
methods to visualize diverse biological data like genomics or transcriptomics
data. They are able to chose suitable visualizations based on the type of data
and the given analysis task. The students will be able to design and develop
complex, interactive visual analytics applications in small teams.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Lecture L o 2 4 O 30 g 100
Tutorials T o 2 2
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for -
participation
Lecturer Krone, Nieselt
Literature Lecture slides will be provided for download. Tamara Munzner ‘Visualization
Analysis and Design’, A K Peters, 2014. Nature Methods Supplement ‘Visual-
izing biological data’, various Nature Methods ‘Points of View’ articles.
Module handbook M.Sc. Machine Learning 56

Module Number: Module title Module

BIO-4331 Advances in Computational Transcriptomics Lecture, Tutorials
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture, Tutorials
Duration 1 semester
Frequency once a year
Language of English
instruction
Type of Exam Written exam (oral exam for small number of participants)
Content Functional genomics, i.e. the interpretation of a genome to determine the bi-
ological function of genes and gene interactions, is one of the most important
fields in modern biology. Today, "next-generation" sequencing technologies are
increasingly being used to measure the expression of thousands of genes simulta-
neously. This results in new challenges for bioinformatics, both algorithmically
and software-wise. In the lecture the following topics will be discussed among
others: NGS technologies, in particular RNA-Seq and ChIP-Seq technologies,
fast to ultrafast alignment methods of short reads, mapping-based and de novo
’assembly’ of genomes and transcriptomes, peak calling, splicing and gene mod-
els, motif search, differential expression, visualization of NGS data and other
current topics. In the exercises, especially scientific work and scientific writ-
ing is encouraged. The exercises are also supplemented with blended learning
methods
Objectives The students are familiar with the new bioinformatics findings on expression
analysis and the newer sequencing technologies. They can formulate the chal-
lenges of the new technologies for bioinformatics. They know algorithms for
the quantification of expression data, statistical methods and machine learning
procedures for the calculation of differential expression and classification as well
as methods for the analysis of expression data in a network context. Students
can analyse real microarray experiments as well as RNA-Seq experiments and
have deepened their R knowledge. The students are aware of the possibilities
but also the limitations of different methods in this subfield of bioinformatics.
They are able to analyse problems on a scientific level and summarise them
in writing. In particular, a high degree of intrinsic motivation and personal
responsibility is encouraged.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Vorlesung V o 2 4 K 90 b 100
Übung Ü o 2 2
Usability (modules) General Computer Science; Expanded Perspectives
Requirement for BIOINF3331 recommended
participation
Lecturer Nieselt
Module handbook M.Sc. Machine Learning 57

(still BIO-4331)
Literature Own lecture notes and selected articles
Module handbook M.Sc. Machine Learning 58

Module Number: Module title Module

BIO-4210 Practical Transcriptomics Practical course
ECTS 3
Work load
- Contact time Work load Class time Self-Study
- Self study 90 h 30 h / 2 CH 60 h
Lecture type Practical course
Duration 1 semester
Frequency Offered at irregular intervals
Language of English
instruction
Type of Exam The final grade is based on performance, a written report on each day of the
practical course, and one or two short oral presentations.
Content The focus is on the practical analysis of so-called next generation sequencing
data. Students learn the use of tools for evaluating this data. This practi-
cal course uses real-life data; the focus is on the entire process of evaluating
experimental data, from quality analyses to in-depth statistical analyses; vari-
ous methods are compared. Topics include de-novo assembly, expression count
calculation, normalization and clustering, machine learning methods and their
application to expression data, statistical methods for calculating differential
expressions, visualization methods, and enrichment methods.
Objectives Students will gain practical experience in designing and programming bioin-
formatics software for analyzing NGS data. They will be able to use libraries
and frameworks, and will acquire knowledge or extend their knowledge of Java
or C++ and R. By working together in groups, students obtain teamwork and
collaboration skills, and they will learn about project organization and pre-
sentation techniques. Students will know about the strengths and weaknesses
and about the limitations of various methods for evaluating high-throughput
transcriptomic data, and will be able to describe and evaluate these methods.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Requirement for
Evaluation

Credit Points /
Status

Grade
CH

CP

Practical course P c 3 3 H.R b 100

Usability (modules) -
Requirement for BIOINF4110, BIOINF4120, BIOINF4331 Advances in Computational Tran-
participation scriptomics (recommended), BIOINF3331 Expression Bioinformatik (recom-
mended)
Lecturer Nieselt
Literature Will be provided at the beginning of the course, if necessary.
Module handbook M.Sc. Machine Learning 59

Module Number: Module title Module

BIO-4363 RNA Bioinformatics Seminar
ECTS 3
Work load
- Contact time Work load Class time Self-Study
- Self study 90 h 30 h / 2 CH 60 h
Lecture type Seminar
Duration 1 semester
Frequency irregularly
Language of English
instruction
Type of Exam Presentation (about 30 minutes) and written elaboration (approx. 10 pages),
leading the discussion once
Content In this seminar, current topics related to computer-aided RNA bioinformatics
will be discussed. These can be, among others, the following: Folding: RNA
structure, thermodynamics, basic folding; RNA Abstract shapes; Compara-
tive Structure Prediction: structure comparison, alignment folding, consen-
sus shapes; Structure Comparison: structure metrics, tree alignment, multiple
structure alignment; RNA gene prediction: prediction from models, predic-
tion from folding, prediction from comparisons; miRNAs: miRNA prediction,
miRNA target prediction; Stochastic Models: HMMs, SCFGs, model train-
ing; 3D-Modelling; Cofolding; RNA Motifs and other topics supplemented by
current research.
Objectives The students can independently work with supervision on a challenging topic
through systematic research. Students gain experience in giving a technical
presentation and producing a technical writeup in bioinformatics. They sum-
marize, assess, classify, scientifically correctly represent and present concepts
and methods of bioinformatic RNA biology. On the one hand, the students will
get an overview of modern knowledge in the field of bioinformatic RNA biology
and thus the importance of this subfield of bioinformatics. On the other hand,
they will know that there are still many open research questions in this field.
By studying current articles, the students have not only improved their reading
and learning skills, but also their personal responsibility. The form of learning
used in the seminar is intended to help the students to develop self-confidence
(presentation) and the ability to criticise and communicate (subsequent discus-
sion).
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Seminar S o 2 3 R 45 b 100
Seminar S o H
Usability (modules)
Requirement for -
participation
Lecturer Nieselt
Literature Articles / scientific publications for each individual topic
Module handbook M.Sc. Machine Learning 60

Module Number: Module title Module

MEDZ-4991 Medical Data Science Lecture, Tutorial
ECTS 6
Work load
- Contact time Work load Class time Self-Study
- Self study 180 h 60 h / 4 CH 120 h
Lecture type Lecture, Tutorial
Duration 1 semester
Frequency once per year
Language of English
instruction
Type of Exam Written Exam
Content This lecture comprises different areas of Medical Data Science. Data Science or
statistical machine learning methods have the potential to transform personal
health care over the coming years. Advances in the technologies have generated
large biological data sets. In order to gain insights that can then be used to
improve preventive care or treatment of patients, these big data have to be
stored in a way that enables fast querying of relevant characteristics of the data
and consequently building statistical models that represent the dependencies
between variables. These models can then be utilized to derive new biomedical
principals, provide evidence for or against certain hypotheses, and to assist
medical professionals in their decision process. Specific topics are:
• Gaining new insights from medical data
• Modeling uncertainty in medical data science models
• Making medical findings available through interpretable decision support
systems
Method-wise, the lecture introduces methods for GWAS analyses (e.g., LMMs),
methods for sequence analysis (e.g., kernel methods), methods for “small n
problems” (e.g., domain adaptation, transfer learning, and multitask learn-
ing), methods for data integration (advanced unsupervised learning methods),
methods for learning probabilistic Machine Learning models (e.g., graphical
models), methods for large data sets (e.g., deep learning models).

Objectives The students are capable of explaining the most important terms, methods and
theories in the data science area with focus on the analysis of biomedical data.
They are enabled to decide which type of methods fit to which kind of data sets.
The students can critically reflect on shortcomings of state-of-the-art methods
to potentially come up with ideas for extending or improving the methods.
Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation

Requirement for
Credit Points /
Status

Grade
CH

CP

Vorlesung V o 2 4 K 90 b 100
Übung Ü o 2 2
Usability (modules) General Computer Science; Expanded Perspectives
Module handbook M.Sc. Machine Learning 61

(still MEDZ-4991)
Requirement for recommended: Machine learning: theory and algorithms or Introduction to
participation Statistical Machine Learning for Bioinfos and Medicine Infos
Lecturer Pfeifer
Literature Trevor Hastie, Robert Tibshirani, Jerome Friedman: The Elements of Statis-
tical Learning, Springer Series in Statistics.
Further books will be announced in the first lecture.
Study Area: Expanded Perspectives

Module Number: Module title Module

ML-5001 Expanded Perspectives Lecture, Tutorial, Semi-
nars
ECTS 12
Work load
- Contact time Work load Class time Self-Study
- Self study 360 h 120 h / 8 CH 240 h
Lecture type Lecture, Tutorial, Seminars
Duration 3 semester
Frequency every semester
Language of English
instruction
Type of Exam Oral or written exams, presentation, essays, reports
Content In this study area, students can choose courses freely from almost all courses
(except for sports courses) offered at the University of Tübingen. In particular
also all courses offered in the area of ‘diverse topics in machine learning’ or
‘general computer science’ can be taken. It is also meant to give students
the opportunity to learn about particular application fields (e.g., geoscience,
linguistics), improve their language skills in German (for foreign students) or
English (for German students), or learn to reflect upon ethical or philosophical
challenges brought by machine learning. Altogether 12 CPs in this field have to
be fulfilled. Courses taken in this area need to be graded ones, and the grades
will show up on the transcript of records, but the grades will not be taken into
account for the cumulative grade of the Master’s program, as stated above.
Due to the high, interdisciplinary flexibility of the courses that can be taken in
this study area, the expected performance in the respective courses are checked
separately, depending on the format.
Objectives These depend on the format and content of the courses taken.
Usability (modules)
Requirement for -
participation
Lecturer -
Literature -

62
Module Master Thesis

Module Number: Module title Module

ML-4999 Master thesis Independent research
work, Master’s thesis (in
written form) and oral
presentation
ECTS 30
Work load
- Contact time Work load Class time Self-Study
- Self study 900 h 30 h / 2 CH 870 h
Lecture type Independent research work, Master’s thesis (in written form) and oral presen-
tation
Duration 1 semester
Frequency Every semester
Language of English
instruction
Type of Exam Written thesis and oral presentation
Content The Master’s thesis is the final stage of the Master’s degree program, and
comprises completing a project in machine learning, evaluating and processing
the results obtained, and finally preparing a written detailed presentation of
these results. The results should be of scientific value. In addition, students
will give an oral presentation of their thesis’ topic.
Objectives Students
• are able to become familiar with a current research issue within a given
time frame. They are able to apply scientific methods and present their
results in a scientifically appropriate manner;
• are able to independently handle a complex scientific issue, applying their
knowledge of machine learning methods;
• gain a deeper understanding of how to solve problems, and are able to
apply their knowledge of methods;
• are able to work in teams in an international scientific setting;
• are able to present and defend their evidence before an audience in En-
glish.

63
Module handbook M.Sc. Machine Learning 64

(still ML-4999)

Duration of Exam
Type of Exam

Calculation of
Type of Class

Module (%)
Evaluation
Requirement for
Credit Points /

Status
Grade

CH

CP
Master’s thesis R o – 27 w g 100
Oral presentation - o – 3
Usability (modules) -
Requirement for If any conditions have been set for admission to a Master’s degree course,
participation students must prove that these conditions have been met prior to registering a
thesis topic.
Lecturer Lecturers of the Department of Computer Science
Literature Depends on the topic