Winning Ant Wars: Evolving a Human-Competitive Game Strategy using Fitnessless Selection

by Wojciech Jaśkowski, Krzysztof Krawiec, Bartosz Wieloch
Abstract:
We tell the story of BrilliAnt, the winner of the Ant Wars contest organized within GECCO’2007, Genetic and Evolutionary Computation Conference. The task for the Ant Wars contestants was to evolve a controller for a virtual ant that collects food in a square toroidal grid environment in the presence of a competing ant. BrilliAnt, submitted to the contest by our team, has been evolved through competitive one-population coevolution using genetic programming and a novel fitnessless selection method. In the paper, we detail the evolutionary setup that lead to BrilliAnt’s emergence, assess its human-competitiveness, and describe selected behavioral patterns observed in its strategy.
Reference:
Winning Ant Wars: Evolving a Human-Competitive Game Strategy using Fitnessless Selection (Wojciech Jaśkowski, Krzysztof Krawiec, Bartosz Wieloch), In Genetic Programming 11th European Conference, EuroGP 2008, Proceedings (M. O’Neill, ed.), Springer-Verlag, volume 4971, 2008.
Bibtex Entry:
@InProceedings{Jaskowski2008antwars,
  Title                    = {Winning {Ant Wars}: Evolving a Human-Competitive Game Strategy using Fitnessless Selection},
  Author                   = {Wojciech Jaśkowski and Krzysztof Krawiec and Bartosz Wieloch},
  Booktitle                = {Genetic Programming 11th European Conference, EuroGP 2008, Proceedings},
  Year                     = {2008},
  Editor                   = {M. O'Neill},
  Month                    = {mar},
  Pages                    = {13--24},
  Publisher                = {Springer-Verlag},
  Series                   = {Lecture Notes in Computer Science},
  Volume                   = {4971},

  Abstract                 = {We tell the story of BrilliAnt, the winner of the Ant Wars contest organized within GECCO'2007, Genetic and Evolutionary Computation Conference. The task for the Ant Wars contestants was to evolve a controller for a virtual ant that collects food in a square toroidal grid environment in the presence of a competing ant. BrilliAnt, submitted to the contest by our team, has been evolved through competitive one-population coevolution using genetic programming and a novel fitnessless selection method. In the paper, we detail the evolutionary setup that lead to BrilliAnt's emergence, assess its human-competitiveness, and describe selected behavioral patterns observed in its strategy.},
  Doi                      = {10.1007/978-3-540-78671-9_2},
  Homepage-url             = {http://www.cs.put.poznan.pl/wjaskowski},
  Keywords                 = {evolutionary computation, genetic programming, evolutionary algorithms, games, coevolution, artificial intelligence},
  Url                      = {http://www.cs.put.poznan.pl/wjaskowski/pub/papers/jaskowski08EuroGPantwars.pdf}
}

This entry was posted by . Bookmark the permalink.