Evolution of Hive Intelligence Using Genetic Programming

Created by W.Langdon from gp-bibliography.bib Revision:1.3973

@Unpublished{widland:1999:ehiGP,
  author =       "Tom Widland and Kevin Oishi and Alex Feuchter and 
                 Ryan Duryea and Ryan Davies",
  title =        "Evolution of Hive Intelligence Using Genetic
                 Programming",
  note =         "WWW pages",
  year =         "1999",
  email =        "Tom Widland ",
  keywords =     "genetic algorithms, genetic programming",
  URL =          "http://www.challenge.nm.org/archive/98-99/finalreports/006/",
  abstract =     "Executive Summary: Our project deals with the
                 evolution of hive intelligence using genetic
                 programming with the classic video game Pacman as our
                 model environment. Pacman is an arcade game where a
                 group of {"}ghosts{"} try to catch a Pacman as he
                 attempts to eat all the dots in a maze in order to
                 progress to the next level. Hive intelligence is the
                 concept that a group of individual organisms working
                 together as a cohesive unit can efficiently accomplish
                 a defined task. In our model of Pacman, the ghosts are
                 the individual organisms that are assigned the task of
                 catching Pacman in a maze as quickly as possible. They
                 work together as a team, communicating with each other
                 to catch the Pacmen. At the end of each simulation our
                 program rates them on a fitness scale to determine
                 their prowess as a team. The ghost team that catches
                 the most Pacmen in a specified amount of time gets the
                 highest fitness score. We take the fittest teams and
                 mix their programs (genes) together using a crossover
                 algorithm. We then run another series of simulations
                 and our program tests the fitness of the new generation
                 of ghost teams. Our results show that genetic
                 programming is a powerful means of evolving a routine
                 to be more effective then any human created algorithm.
                 The applications of such a process are staggering. In
                 almost any situation in which computer programs are
                 used to perform a single, definable task in varying
                 situations, genetic programming can be used to increase
                 the efficiency of the program. From simulating the
                 function of organs in the human body to the exploration
                 of planets, genetic programming is a useful tool in
                 creating the best routines for the job.",
  notes =        "Code pacman.cpp etc in allfiles.tgz",
}

Genetic Programming entries for Tom Widland Kevin Oishi Alex Feuchter Ryan Duryea Ryan Davies

Citations