ProfileCopyright: Peter Winandy
Uncertainty is nowadays more and more pervasive in computer science. It is important both in big data and at the level of events and control. Applications have to treat lots of data, often from unreliable sources such as noisy sensors and untrusted web pages. Data may also be subject to continuous changes, may come in different formats, and is often incomplete.
Systems have to deal with unpredictable and sometimes hostile environments. A different, also inevitable, kind of uncertainty arises from abstractions in system models focusing on the control of events. Probabilistic modelling and randomization are key techniques for dealing with uncertainty. Many trends witness this. Real-world modelling in planning is advancing by probabilistic programs describing complex Bayesian networks. In security, hostile environments are often captured by probabilistic adversaries. Probabilistic databases deal with uncertain data by associating probabilities to the possible worlds. In systems verification, probabilistic model checking has emerged as a key technique allowing for correctness checking and performance analysis. Similar developments take place in logic and game theory.
The pervasiveness of uncertainty urges to make substantial enhancements in probabilistic modelling and reasoning so as to understand, reason about, and master uncertainty. The focus of the interdisciplinary Research Training Group UnRAVeL is to significantly advance probabilistic modelling and analysis for uncertainty by developing new theories, algorithms, and tool-supported verification techniques, and to apply them to core problems from security, planning , and safety and performance analysis. To tackle these research challenges, theoretical computer scientists from computer-aided verification, logic and games, algorithms and complexity, together with experts from management science , and railway engineering form the core of this Research Training Group.
The qualification and supervision concept aims at offering the Ph.D. students an optimal environment to carry out their research. Every Ph.D. student has two supervisors; the rights and duties of the supervisors and students are laid down in a written supervision agreement. The curriculum consists of bi-weekly research seminars, soft-skill courses, reading groups, annual workshops, a summer school in the first Ph.D. year, and advanced guest-lectures.