Application of reinforcement learning algorithms to software verification

Authors: Moturu, Krishna Priya Darsini
Advisor: Laviolette, François
Abstract: This thesis presents a novel form of system verification through reinforcement learning algorithms. Large and complex software systems are often developed as a team effort. The aim of the development is to satisfy the customer by delivering the right product, with the right quality, and in time. Errors made by developers will always occur when a system is developed, but their effect can be reduced by removing them as early as possible. Software verification and validation are activities that are conducted to im- prove product quality. In this thesis we will adapt the techniques used in reinforcement learning to Software verification to verify if implemented system meets its specifica- tions. This new approach can be used even if the complete model of the system is not available, which is new in probabilistic verification. This thesis main aim is not only to answer the question whether the system behaves according to its specifications but also to find the degree of divergence between the system and its specifications.
Document Type: Mémoire de maîtrise
Issue Date: 2006
Open Access Date: 12 April 2018
Grantor: Université Laval
Collection:Thèses et mémoires

Files in this item:
23583.pdf804.64 kBAdobe PDFView/Open
All documents in CorpusUL are protected by Copyright Act of Canada.