Active Learning on Visibly One Counter Systems



This report focuses on an active learning algorithm for the class of visibly one-counter automata,and its implementation in python and C++. The algorithm allows creating such automata using queriesto a teacher, with no knowledge on the final automaton. We then discuss the possible improvements of the algorithm regarding the algorithm execution speed and the automaton represention. We also give an overview of how benchmarks could be perform on such a semi-automatic program.