Improving the quality of supervised finite-state machine construction using real-valued variables

Abstract

The use of finite-state machines (FSMs) is a reliable choice for control system design since they can be formally verified. In this paper a problem of constructing FSMs with real-valued input and control parameters is considered. It is supposed that a set of human-created behavior examples, or tests, is available. One of the earlier approaches for solving the problem suggested using genetic algorithms together with a transition labeling algorithm. This paper improves this approach via the use of real-valued variables which are calculated using the FSM's input data. FSMs with realvalued variables are represented as systems of linear controllers. The new approach allows to synthesize FSMs of better quality than it was possible earlier.

Measuring the Quality of B Abstract Machines with ISO/IEC 25010

“…(18) for learnability. In automatic model construction, model quality can be measured using fitness functions that reflect the distance between state transition systems and observations in the real world [10]. The fitness functions are similar to PFComp and PFCorr because they concern whether a model can fit functional requirements.…”