A model learning based testing approach for spiking neural P systems

Abstract

This paper proposes a testing approach for spiking neural P systems, significantly different from the past testing research for cell-like P systems. The proposed method provides a solution to the state explosion problem by constructing a series of approximations, using the concept of cover automaton and Angluin-style model learning from queries, more precisely the Ll algorithm for learning a finite cover automaton, adapted to the more general X-machine model. Furthermore, the concept of identifiability, which is an essential prerequisite for the successful application of our method, but also a more general design characteristic inspired from the testing practice, is introduced and investigated in the context of spiking neural P systems. Identifiability of system's components (modules, methods, etc.) is a fundamental criterion used for assessing a system's testability since it allows the components of a system to be identified from the behaviour produced in response to the inputs received and, consequently, maximizes the effectiveness of the testing process.