Gen-O-Fix: An embeddable framework for Dynamic Adaptive Genetic Improvement Programming

Citation

Swan J, Epitropakis M & Woodward J (2014) Gen-O-Fix: An embeddable framework for Dynamic Adaptive Genetic Improvement Programming. Department of Computing Science and Mathematics Technical Report, CSM-195. University of Stirling. http://www.cs.stir.ac.uk/~jsw/genofix-TR.pdf

Abstract
Genetic Improvement Programming (GIP) is concerned with automating the burden of software maintenance, the most costly phase of the software lifecycle. We describe Gen-O-Fix, a GIP frame- work which allows a software system hosted on the Java Virtual Machine to be continually improved (e.g. make better predictions; pass more regression tests; reduce power consumption). It is the first exemplar of a dynamic adaptive GIP framework, i.e. it can improve a system as it runs. It is written in the Scala programming language and uses reflection to yield source-to-source transforma- tion. One of the design goals for Gen-O-Fix was to create a tool that is user-centric rather than researcher-centric: the end-user is required only to provide a measure of system quality and the URL of the source code to be improved. We discuss potential applications to predictive, embedded and high-performance systems.