Improving Generalization of Evolved Programs Through Automatic Simplification

Created by W.Langdon from gp-bibliography.bib Revision:1.3989

@InProceedings{Helmuth:2017:GECCO,
  author =       "Thomas Helmuth and Nicholas Freitag McPhee and 
                 Edward Pantridge and Lee Spector",
  title =        "Improving Generalization of Evolved Programs Through
                 Automatic Simplification",
  booktitle =    "Proceedings of the Genetic and Evolutionary
                 Computation Conference",
  series =       "GECCO '17",
  year =         "2017",
  isbn13 =       "978-1-4503-4920-8",
  address =      "Berlin, Germany",
  pages =        "937--944",
  size =         "8 pages",
  URL =          "http://doi.acm.org/10.1145/3071178.3071330",
  DOI =          "doi:10.1145/3071178.3071330",
  acmid =        "3071330",
  publisher =    "ACM",
  publisher_address = "New York, NY, USA",
  keywords =     "genetic algorithms, genetic programming, automatic
                 simplification, generalization, overfitting, push",
  month =        "15-19 " # jul,
  abstract =     "Programs evolved by genetic programming unfortunately
                 often do not generalize to unseen data. Reliable
                 synthesis of programs that generalize to unseen data is
                 therefore an important open problem. We present
                 evidence that smaller programs evolved using the PushGP
                 system tend to generalize better over a range of
                 program synthesis problems. Like in many genetic
                 programming systems, programs evolved by PushGP usually
                 have pieces that can be removed without changing the
                 behaviour of the program. We describe methods for
                 automatically simplifying evolved programs to make them
                 smaller and potentially improve their generalization.
                 We present five simplification methods and analyse
                 their strengths and weaknesses on a suite of general
                 program synthesis benchmark problems. All of our
                 methods use a straightforward hill-climbing procedure
                 to remove pieces of a program while ensuring that the
                 resulting program gives the same errors on the training
                 data as did the original program. We show that
                 automatic simplification, previously used both for
                 post-run analysis and as a genetic operator, can
                 significantly improve the generalization rates of
                 evolved programs.",
  notes =        "Also known as \cite{Helmuth:2017:IGE:3071178.3071330}
                 GECCO-2017 A Recombination of the 26th International
                 Conference on Genetic Algorithms (ICGA-2017) and the
                 22nd Annual Genetic Programming Conference (GP-2017)",
}

Genetic Programming entries for Thomas Helmuth Nicholas Freitag McPhee Edward R Pantridge Lee Spector

Citations