Signal Processing and Machine Learning

Module Number: EI70380

Duration: 1 Semester

Ocurrence: Summer Semester

Language: English

Number of ECTS: 5


Professor in charge: Wolfgang Utschick

Amount of work

Contact hours: 60

Self-study hours: 90

Total: 150

Description of achievement and assessment methods

A written examination (90 min) assesses the students' abilities to evaluate basic and advanced concepts of signal processing and machine learning in typical applications in information and communication technology. The examination consists of calculations and short questions about problems in the field of Signal Processing and Machine Learning. The exam is closed-book. As supporting material, it is allowed to use up to 10 DIN-A4 sheets with arbitrary (handwritten or printed) notes. The use of electronic devices such as calculators, cell phones, notebooks, and similar devices is not allowed. For students whose mother tongue is not English, a dictionary English-mother tongue in print form is allowed as long as no handwritten notes are in it.

Recommended requirements

Linear Algebra and Calculus, Statistical Signal Processing, Convex Optimization

Intended Learning Outcomes

After successfully passing the module, the students are able to understand, apply, evaluate, and create mathematical concepts and numerical algorithms in the field of signal processing and machine learning for communications and data processing applications. Furthermore, the students will be able to reformulate typical problem formulations in order to apply sparse signal processing techniques and machine learning algorithms and have gained insight into current cutting-edge research problems in these fields.


Introduction of advanced mathematical methods, concepts, and algorithms for selected topics in signal processing and machine learning and their application in current cutting-edge research in communications and data processing applications, which highlights a joint perspective on both paradigms. Introduction into the basics of estimation and classification theory, support vector machine and kernel methods, random forests, neural networks, deep neural networks, recurrent neural networks, sparse signal processing and compressive sensing for machine learning. The usage of popular toolboxes will be demonstrated in selected application examples. The curriculum may change in any semester and will be announced in time.

Teaching and Learning Methods

Mathematical concepts and numerical algorithms for selected topics in signal processing and machine learning are introduced during the lectures. They are transferred by means of case studies and applications which demonstrate the use of the introduced concepts and their respective numerical algorithms. The students further investigate the introduced concepts by solving specific problem formulations and by applying and programming own numerical algorithms and available toolboxes.


The following kinds of media are used:

- Presentations.

- Lecture notes.

- Exercises with solutions as download.


There is no general recommendation of literature because of the widespread field of potential topics and applications. Literature relevant for the covered topics will be recommended in the course of the semester.