Using features of models to improve state space exploration

Heijblom, A.R. (2016) Using features of models to improve state space exploration.

[img]
Preview
PDF
1MB
Abstract:State space methods are a popular approach to perform formal verification. In the past decades many methods did arise to cope with the state spaces of larger models. Due to the wide variety of strategies and models is may be hard for a user to select an appropriate strategy. If a bad strategy is selected, the given model can be unsolvable or the process may waste resources like time and memory. Moreover, the intervention of the user makes state space methods less automated. Therefore, it would be convenient if model checking tools itself determine the strategy for a given model. This process requires model checking tools to predict a strategy based on the information presented in a given model. Our research investigates to what extent characteristics of a model can be used to predict an appropriate strategy. The performance of 784 different PNML and DVE models was determined using LTSmin for 60 selected strategies. This information was used to create several classifiers using machine learning techniques. The classifiers should predict an appropriate strategy given eleven selected features of a model.
Item Type:Essay (Master)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Computer Science MSc (60300)
Link to this item:http://purl.utwente.nl/essays/71574
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page