Opaque models belonging to the machine learning world are ever more exploited
in the most different application areas. These models, acting as black boxes
(BB) from the human perspective, cannot be entirely trusted if the application
is critical unless there exists a method to extract