The Communications of ACM published an article on automata learning in software engineering last February. The techniques described in the article are used to obtain models for the (input/output) behaviour of software. Even without access to source code, one can now use model checking or other bug finding tools on these models. The article shows many successful applications. Why is this possible at all?

Continue Reading "Why is learning so effective in software testing?"

This year a team from the Radboud University and TU Delft used automata learning to compete in the RERS challenge 2016. The challenge provides (generated) source code where the challenge is to (dis)prove certain LTL formulas and to analyze which error states are reachable. Information on this challenge can be found here: Commonly, learning is not used in this competition and only white-box methods are used. This year, however, automata learning was applied to great success. For the problems where LTL formulas had to be…Continue Reading “Succes at the RERS challenge 2016 for automata learning”