$N$-version Genetic Programming via Fault Masking

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

  title =        "{$N$}-version Genetic Programming via Fault Masking",
  author =       "Kosuke Imamura and Robert B. Heckendorn and 
                 Terence Soule and James A. Foster",
  editor =       "James A. Foster and Evelyne Lutton and 
                 Julian Miller and Conor Ryan and Andrea G. B. Tettamanzi",
  booktitle =    "Genetic Programming, Proceedings of the 5th European
                 Conference, EuroGP 2002",
  volume =       "2278",
  series =       "LNCS",
  pages =        "172--181",
  publisher =    "Springer-Verlag",
  address =      "Kinsale, Ireland",
  publisher_address = "Berlin",
  month =        "3-5 " # apr,
  year =         "2002",
  keywords =     "genetic algorithms, genetic programming",
  ISBN =         "3-540-43378-3",
  DOI =          "doi:10.1007/3-540-45984-7_17",
  abstract =     "We introduce a new method, N-Version Genetic
                 Programming (NVGP), for building fault tolerant
                 software by building an ensemble of automatically
                 generated modules in such a way as to maximize their
                 collective fault masking ability. The ensemble itself
                 is an example of n-version modular redundancy for fault
                 tolerance, where the output of the ensemble is the most
                 frequent output of n independent modules. By maximising
                 collective fault masking, NVGP approaches the fault
                 tolerance expected from n version modular redundancy
                 with independent faults in component modules. The
                 ensemble comprises individual modules from a large pool
                 generated with genetic programming, using operators
                 that increase the diversity of the population. Our
                 experimental test problem classified promoter regions
                 in Escherichia coli DNA sequences. For this problem,
                 NVGP reduced the number and variance of errors over
                 single modules produced by GP, with statistical
  notes =        "EuroGP'2002, part of \cite{lutton:2002:GP}, UCI ML
                 e.coli benchmark (balanced training 35 positives, 35
                 negatives). beowulf. 2-gram (16 possible). linear gp
                 (MIPS like). max length 80. 4 read/write registers
                 (memory). 5 crossover types. Inversion (!). 2 mutation
                 operators, tournament fitness=correlation coefficient.
                 40 isolated islands (demes) each 100 individuals.
                 ensemble = composition from (randomly chosen)

                 ensemble is qualified if number of its errors <= number
                 of errors expected if its components were _independent_
                 14% to 58% improvement in error rate for ensemble (of
                 30) compared to single GP (pop 100).


Genetic Programming entries for Kosuke Imamura Robert B Heckendorn Terence Soule James A Foster