Group colloquium: Learning-Based Testing in the LearnLib

When: Feb. 7, 2019, 15:45-16:30

Where: Ravelijn 2503

Who: Jeroen Meijer

Learning-based testing (or black-box checking) is a process that combines active automata learning with model checking.
Active automata learning algorithms are used to infer the behavior of a particular system under test in the form of finite state machines. These learned FSMs are hypothetical abstractions of the system's behavior. Since the FSMs are only hypothetical, applying model checking to the FSMs can have two results. Either the hypothesis is incorrect, and it can be refined, or the system does not satisfy the property that is checked. To this end, we will highlight the active automata learning algorithm and how it is augmented with both safety and liveness properties expressed in LTL. By using the RERS challenge we will show which algorithms in the LearnLib perform the best in the number of so called queries posed to the system.