Towards an Information Theoretic Framework for Genetic programming

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

@InCollection{Card:2007:GPTP,
  author =       "Stuart W. Card and Chilukuri K. Mohan",
  title =        "Towards an Information Theoretic Framework for Genetic
                 programming",
  booktitle =    "Genetic Programming Theory and Practice {V}",
  year =         "2007",
  editor =       "Rick L. Riolo and Terence Soule and Bill Worzel",
  series =       "Genetic and Evolutionary Computation",
  chapter =      "6",
  pages =        "87--106",
  address =      "Ann Arbor",
  month =        "17-19 " # may,
  publisher =    "Springer",
  keywords =     "genetic algorithms, genetic programming",
  DOI =          "doi:10.1007/978-0-387-76308-8_6",
  size =         "19 pages",
  abstract =     "An information-theoretic framework is presented for
                 the development and analysis of the ensemble learning
                 approach of genetic programming. As evolution proceeds,
                 this approach suggests that the mutual information
                 between the target and models should: (i) not decrease
                 in the population; (ii) concentrate in fewer
                 individuals; and (iii) be distilled from the inputs,
                 eliminating excess entropy. Normalised information
                 theoretic indices are developed to measure fitness and
                 diversity of ensembles, without a priori knowledge of
                 how the multiple constituent models might be composed
                 into a single model. With the use of these indexes for
                 reproductive and survival selection, building blocks
                 are less likely to be lost and more likely to be
                 recombined. Price's Theorem is generalised to pair
                 selection, from which it follows that the heritability
                 of information should be stronger than the heritability
                 of error, improving evolvability. We support these
                 arguments with simulations using a logic function
                 benchmark and a time series application. For a chaotic
                 time series prediction problem, for instance, the
                 proposed approach avoids familiar difficulties
                 (premature convergence, deception, poor scaling, and
                 early loss of needed building blocks) with standard GP
                 symbolic regression systems; information-based fitness
                 functions showed strong intergenerational correlations
                 as required by Price's Theorem.",
  notes =        "http://www.cscs.umich.edu/events/gptp2007/

                 Card-Mohan-draft-2007-4-4.pdf

                 part of \cite{Riolo:2007:GPTP} To be published after
                 workshop Jan 2008?",
}

Genetic Programming entries for Stu Card Chilukuri K Mohan

Citations