Der Inhalt dieser Seite ist aktuell nur auf Englisch verfügbar.

Starting with the second year of the master programme Applied Image and Signal Processing, specific application scenarios are discussed and corresponding technologies are investigated in a number of elective courses. Choose two such courses from the list on the next page and complement those by free electives with a total sum of 6 ECTS in the 3rd or 4th semester. While it is recommended to take a third course from the aforementioned list, other lectures given in English on one of the two universities also qualify as free electives.

Electives

Natural Language Processing

at the Salzburg University of Applied Sciences
Attention-based models have become an indespensible part of modern natural language processing applications, e.g. in the field of social media analysis or human-machine interfaces. You will design and implement appropriate models for areas such as machine translation and sentiment analysis using contextualized text representations and complex neural network architectures. We also take a look at dialog systems and language generation for conversational purposes.

Reinforcement Learning

at the Salzburg University of Applied Sciences
Reinforcement Learning (RL) is a learning paradigm to train intelligent agents to optimize their actions in order to maximize a reward given by an environment. It is vital to smart solutions in areas such as robotics, industry 4.0, trading or gaming. Students will identify problems suited for RL, find appropriate models and assemble solutions using toolboxes. Special attention is paid to upcoming areas such as Deep RL and model-based RL that address known issues in real-world applications of RL.

Medical Imaging

at the Paris Lodron University Salzburg
Image and signal processing applications in medicine are optimized with respect to the numerous modalities and sensors used. Images need to be segmented, co-registered and processed with respect to contextual knowledge. You get to know the most popular tools and libraries, and acquire competences in designing solutions for tomorrow‘s medical technology.

Biometric Systems

at the Paris Lodron University Salzburg
Study biometric technologies and learn the most common modalities for the identification of individuals. Implications on security and privacy are discussed. Biometry is a generic topic in that various methods and concepts can be found in many other areas of image and signal processing as well, and in that the optimization of systems with respect to risk minimization is of a general nature.

Media Security

at the Paris Lodron University Salzburg
Concepts for encryption, authentification and robust labelling of multimedia data are presented and their application in media forensics is discussed. Since different modalities require highly specialized methods, you will become an expert in a range of such algorithms and will be able to design applications that achieve a good compromise between robustness, speed and usability.

Computational Geometry

at the Paris Lodron University Salzburg
Computational geometry is the study of the design and analysis of efficient algorithms for solving problems with a geometric flavor. The methodologies of computational geometry allow one to investigate solutions of numerous geometric problems that arise in application areas such as image processing, computer-aided design, manufacturing, geographic information systems, robotics and graphics. This course offers an introduction to computational geometrylike geometric searching, convex hulls, Voronoi diagrams, straight skeletons, triangulations, and robustness issues.

Machine Learning

at the Paris Lodron University Salzburg
You study how to program computers to »learn« from available input data. In other words, it is the process of converting experience in the form of training data into expertise to solve a variety of different tasks. Fundamental concepts such as probably approximately correct (PAC) learning, Vapnik–Chervonenkis theory and applications thereof are considered and applied in the analysis of popular learning algorithms such as boosting or support vector machines.