Authors: Merten, Maik
Title: Active automata learning for real life applications
Language (ISO): en
Abstract: Automata learning is a concept discussed in the literature for decades. Accordingly, the theoretical framework for learning automata from observations has been in place already for a considerable time. Despite the ever-increasing theoretical maturity of the field, real-life applications are few and far between. In part this can certainly be attributed to the lack of ready-made infrastructure, e.g., frameworks that support automata learning with the goal of learning realistic systems. Additionally, the degree of automation in this field is low, meaning that learning setups have to be instantiated manually and per-system, making this a time-consuming and laborious undertaking. The central question of this thesis is "How can active automata learning be readied for application on real-life systems?". Contributions presented includes work on learning frameworks and tools, learning algorithms, scalability of learning solutions, and automated configuration and execution of learning setups.
Subject Headings: Automata learning
Learning frameworks
Mealy machines
Regular inference
URI: http://hdl.handle.net/2003/29884
http://dx.doi.org/10.17877/DE290R-5169
Issue Date: 2013-01-29
Appears in Collections:LS 05 Programmiersysteme und Übersetzerbau

Files in This Item:
File Description SizeFormat 
dissertation-pdfa.pdfDNB2.22 MBAdobe PDFView/Open


This item is protected by original copyright



All resources in the repository are protected by copyright.