Learning Mealy Machines with One Timer

Abstract

We present Mealy machines with a single timer (MM1Ts), a class of models that is both sufficiently expressive to describe the realtime behavior of many realistic applications, and can be learned efficiently. We show how learning algorithms for MM1Ts can be obtained via a reduction to the problem of learning Mealy machines. We describe an implementation of an MM1T learner on top of LearnLib, and compare its performance with recent algorithms proposed by Aichernig et al. and An et al. on several realistic benchmarks.

A New Approach for Active Automata Learning Based on Apartness

“…Isberner [36] shows that these descendants of L * can be described in a single, general framework. 1 Variations of L * have also been used as a basis for learning extensions of DFAs such as Mealy machines [48], I/O automata [2], non-deterministic automata [16], alternating automata [6], register automata [1,17], nominal automata [46], symbolic automata [40,7], weighted automata [14,11,30], Mealy machines with timers [64], visibly pushdown automata [36], and categorical generalisations of automata [62,29,12,18]. It is fair to say that L * -like algorithms completely dominate the research area of active automata learning.…”