Mutation-Based Genetic Improvement of Software

Created by W.Langdon from gp-bibliography.bib Revision:1.3872

@PhdThesis{Thesis_Fan_v2.1,
  author =       "Fan Wu",
  title =        "Mutation-Based Genetic Improvement of Software",
  school =       "Department of Computer Science, University College,
                 London",
  year =         "2017",
  address =      "UK",
  month =        jul # " 2",
  keywords =     "genetic algorithms, genetic programming, Genetic
                 Improvement, SBSE, deep parameters, mutation testing",
  URL =          "http://www.cs.ucl.ac.uk/staff/W.Langdon/ftp/papers/Thesis_Fan_v2.1.pdf",
  size =         "144 pages",
  abstract =     "Genetic Improvement (GI) of software is a recent field
                 that has drawn much attention from Software Engineering
                 researchers. It aims to use search techniques to
                 automatically modify and improve existing software. The
                 drawback in previous GI approaches is scalability of
                 these approaches, due to the large search space formed
                 by the code base in real-world systems. To overcome the
                 scalability challenge, more recent studies have
                 confined the granularity of code modification at the
                 statement level and applied a prior sensitivity
                 analysis to further reduce the search space. However,
                 some software improvements may require code changes at
                 a finer level of granularity.

                 This thesis demonstrates that, by combining with
                 Mutation Testing techniques, GI can operate at this
                 finer granularity while preserving scalability. The
                 thesis applies Mutation Operators to automatically
                 modify the source code of the target software. After a
                 prior sensitivity analysis on First Order Mutants, deep
                 (previously unavailable) parameters are exposed from
                 the most sensitive locations, followed by a
                 bi-objective optimisation process to fine tune them
                 together with existing (shallow) parameters. The
                 objective is to improve both time and memory resources
                 required by the computation.

                 Since this approach relies on the selection of Mutation
                 Operators and traditional Mutation Operators are not
                 concerned with memory performance, the thesis proposes
                 and evaluates Memory Mutation Operators in the Mutation
                 Testing context. Using both traditional and Memory
                 Mutation Operators, the thesis further seeks to improve
                 the target software by searching for Higher Order
                 Mutants (HOMs). The thesis presents the result of a
                 code analysis study, which reveals that, among all the
                 code modifications that contribute to the improvement,
                 more than half of them require a finer control of the
                 code, which our approach is better at than previous GI
                 approaches.",
  notes =        "Supervisor: Mark Harman",
}

Genetic Programming entries for Fan Wu

Citations