Vortrag vom 08.01.2014

Vortragender: Christian Schwarz

Titel: Statistical Parameter-Synthesis for Hybrid Systems

Abstract:

Embedded systems are often employed in safety-critical environments. They can be understood as hybrid systems, which feature both: discrete and continuous state transitions. Hybrid automata are an established formal modelling language for hybrid systems. For an underspecified automaton, parameter-synthesis is the task of finding a safe subset of a given parameter space. It is useful during system design and could be used to verify a whole class of systems at once. Classical methods use symbolic execution and variable elimination, and thus can only be applied to small systems with restricted dynamics.
In this talk an alternative approach will be presented. It uses concrete
executions of a probabilistic model, machine learning techniques and hypothesis testing in order to derive a safe parameter set and can be applied to systems with arbitrary dynamics.