Genetic Parallel Programming - Evolving Linear Machine Codes on a Multiple-ALU Processor

Created by W.Langdon from gp-bibliography.bib Revision:1.3963

@InProceedings{Leung:2002:ICAIET,
  author =       "Kwong Sak Leung and Kin Hong Lee and Sin Man Cheang",
  title =        "Genetic Parallel Programming - Evolving Linear Machine
                 Codes on a Multiple-ALU Processor",
  booktitle =    "Proceedings of International Conference on Artificial
                 Intelligence in Engineering and Technology - ICAIET
                 2002",
  year =         "2002",
  editor =       "Sazali Yaacob and R. Nagarajan and Ali Chekima",
  pages =        "207--213",
  month =        "17-18 " # jun,
  address =      "Sabah, Malaysia",
  publisher_address = "Sabah, Malaysia",
  organisation = "School of Engineering and Information Technology,
                 Universiti Malaysia Sabah",
  publisher =    "Universiti Malaysia Sabah",
  keywords =     "genetic algorithms, genetic programming",
  ISBN =         "983-2188-92-X",
  abstract =     "Genetic Programming (GP) is a robust method in
                 Evolutionary Computation. There are two main streams in
                 GP, namely, Tree-based GP (TGP) and Linear GP (LGP).
                 TGP evolves programs represented in tree structure. LGP
                 evolves sequential programs directly. LGP suffers from
                 inflexibility while TGP suffers from inefficiency. This
                 paper proposes a novel framework of an integrated
                 system called Genetic Parallel Programming (GPP) for
                 evolving optimal parallel programs by LGP. The core of
                 the GPP consists of a Multi-ALU Processor (MAP) and an
                 Evolution Engine (EE). The MAP uses Multiple
                 Instruction streams Multiple Data streams (MIMD)
                 architecture. The EE uses a two-phase evolutionary
                 approach and a new GP operation to swap
                 sub-instructions in a parallel program. Three
                 experiments (i.e. Cubic function, Sextic function and
                 Artificial Ant - Santa Fe Trail) are given as examples
                 to show that GPP could discover novel parallel programs
                 that fully use the processor's parallelism. The GPP
                 opens up an entire new opportunity for solving problems
                 with appropriate parallel architecture and learning
                 optimal programs/algorithms automatically.",
  notes =        "http://books.google.co.uk/books?id=ejBaPgAACAAJ",
}

Genetic Programming entries for Kwong-Sak Leung Kin-Hong Lee Ivan Sin Man Cheang

Citations