Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.742336
Title: Automatic selection of statistical model checkers for analysis of biological models
Author: Bakir, Mehmet Emin
ISNI:       0000 0004 7228 3644
Awarding Body: University of Sheffield
Current Institution: University of Sheffield
Date of Award: 2017
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Statistical Model Checking (SMC) blends the speed of simulation with the rigorous analytical capabilities of model checking, and its success has prompted researchers to implement a number of SMC tools whose availability provides flexibility and fine-tuned control over model analysis. However, each tool has its own practical limitations, and different tools have different requirements and performance characteristics. The performance of different tools may also depend on the specific features of the input model or the type of query to be verified. Consequently, choosing the most suitable tool for verifying any given model requires a significant degree of experience, and in most cases, it is challenging to predict the right one. The aim of our research has been to simplify the model checking process for researchers in biological systems modelling by simplifying and rationalising the model selection process. This has been achieved through delivery of the various key contributions listed below. • We have developed a software component for verification of kernel P (kP) system models, using the NuSMV model checker. We integrated it into a larger software platform (www.kpworkbench.org). • We surveyed five popular SMC tools, comparing their modelling languages, external dependencies, expressibility of specification languages, and performance. To best of our knowledge, this is the first known attempt to categorise the performance of SMC tools based on the commonly used property specifications (property patterns) for model checking. • We have proposed a set of model features which can be used for predicting the fastest SMC for biological model verification, and have shown, moreover, that the proposed features both reduce computation time and increase predictive power. • We used machine learning algorithms for predicting the fastest SMC tool for verification of biological models, and have shown that this approach can successfully predict the fastest SMC tool with over 90% accuracy. • We have developed a software tool, SMC Predictor, that predicts the fastest SMC tool for a given model and property query, and have made this freely available to the wider research community (www.smcpredictor.com). Our results show that using our methodology can generate significant savings in the amount of time and resources required for model verification.
Supervisor: Stannett, Mike ; Gheorghe, Marian Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.742336  DOI: Not available
Share: