Improving Source Code with Genetic Programming
Created by W.Langdon from
gp-bibliography.bib Revision:1.8081
- @InProceedings{Langdon:gecco2014_hot,
-
author = "William B. Langdon and Mark Harman",
-
title = "Improving Source Code with Genetic Programming",
-
booktitle = "GECCO 2014 Hot of the Press",
-
year = "2014",
-
editor = "Christian Igel and Dirk Arnold",
-
address = "Vancouver, Canada",
-
publisher_address = "New York, NY, USA",
-
month = "12-15 " # jul,
-
organisation = "SIGEvo",
-
publisher = "ACM",
-
keywords = "genetic algorithms, genetic programming, genetic
improvement, SBSE, gismo",
-
URL = "http://www.cs.ucl.ac.uk/staff/W.Langdon/gismo/Langdon_gecco2014_hot.pdf",
-
size = "1 page",
-
abstract = "Optimising Existing Software with Genetic Programming
(to appear in IEEE Transactions on Evolutionary
Computation doi:10.1109/TEVC.2013.2281544)
\cite{Langdon:2013:ieeeTEC} describes a recent
experiment in which a state-of-the-art Bioinformatics
program was manipulated by GP to give a variant
automatically tuned to a particular task. The program
consists of 50000 lines of C++. Evolution was able to
find a change which speeds it up on the chosen task on
average by a factor of 70, yet still give good answers,
indeed the results are slightly better.
GISMOE uses a BNF grammar specific to Bowtie2 when
mutating genetically improved programs (GIP). Patches
delete, move or insert existing lines of code. No new
code is created. Mutants' fitness is measured by
running them using DNA from The Thousand Genomes
Project.",
-
notes = "Abstract only in conference booklet.
GECCO-2014 A joint meeting of the twenty third
international conference on genetic algorithms
(ICGA-2014) and the nineteenth annual genetic
programming conference (GP-2014)",
- }
Genetic Programming entries for
William B Langdon
Mark Harman
Citations