Constrained novelty search

A study on game content generation

Antonios Liapis, Georgios N. Yannakakis, Julian Togelius

    Research output: Contribution to journalArticle

    Abstract

    Novelty search is a recent algorithm geared toward exploring search spaces without regard to objectives. When the presence of constraints divides a search space into feasible space and infeasible space, interesting implications arise regarding hownovelty search explores such spaces. This paper elaborates on the problem of constrained novelty search and proposes two novelty search algorithms which search within both the feasible and the infeasible space. Inspired by the FI-2pop genetic algorithm, both algorithms maintain and evolve two separate populations, one with feasible and one with infeasible individuals, while each population can use its own selection method. The proposed algorithms are applied to the problem of generating diverse but playable game levels, which is representative of the larger problem of procedural game content generation. Results show that the two-population constrained novelty search methods can create, under certain conditions, larger and more diverse sets of feasible game levels than current methods of novelty search, whether constrained or unconstrained. However, the best algorithm is contingent on the particularities of the search space and the genetic operators used. Additionally, the proposed enhancement of offspring boosting is shown to enhance performance in all cases of two-population novelty search.

    Original languageEnglish (US)
    Pages (from-to)101-129
    Number of pages29
    JournalEvolutionary Computation
    Volume23
    Issue number1
    DOIs
    StatePublished - Mar 16 2015

    Fingerprint

    Game
    Search Space
    Genetic Operators
    Boosting
    Search Methods
    Genetic algorithms
    Search Algorithm
    Divides
    Enhancement
    Genetic Algorithm

    ASJC Scopus subject areas

    • Computational Mathematics

    Cite this

    Constrained novelty search : A study on game content generation. / Liapis, Antonios; Yannakakis, Georgios N.; Togelius, Julian.

    In: Evolutionary Computation, Vol. 23, No. 1, 16.03.2015, p. 101-129.

    Research output: Contribution to journalArticle

    Liapis, Antonios ; Yannakakis, Georgios N. ; Togelius, Julian. / Constrained novelty search : A study on game content generation. In: Evolutionary Computation. 2015 ; Vol. 23, No. 1. pp. 101-129.
    @article{16e79bcf91c54d81ad87d4a29c9dc673,
    title = "Constrained novelty search: A study on game content generation",
    abstract = "Novelty search is a recent algorithm geared toward exploring search spaces without regard to objectives. When the presence of constraints divides a search space into feasible space and infeasible space, interesting implications arise regarding hownovelty search explores such spaces. This paper elaborates on the problem of constrained novelty search and proposes two novelty search algorithms which search within both the feasible and the infeasible space. Inspired by the FI-2pop genetic algorithm, both algorithms maintain and evolve two separate populations, one with feasible and one with infeasible individuals, while each population can use its own selection method. The proposed algorithms are applied to the problem of generating diverse but playable game levels, which is representative of the larger problem of procedural game content generation. Results show that the two-population constrained novelty search methods can create, under certain conditions, larger and more diverse sets of feasible game levels than current methods of novelty search, whether constrained or unconstrained. However, the best algorithm is contingent on the particularities of the search space and the genetic operators used. Additionally, the proposed enhancement of offspring boosting is shown to enhance performance in all cases of two-population novelty search.",
    author = "Antonios Liapis and Yannakakis, {Georgios N.} and Julian Togelius",
    year = "2015",
    month = "3",
    day = "16",
    doi = "10.1162/EVCO_a_00123",
    language = "English (US)",
    volume = "23",
    pages = "101--129",
    journal = "Evolutionary Computation",
    issn = "1063-6560",
    publisher = "MIT Press Journals",
    number = "1",

    }

    TY - JOUR

    T1 - Constrained novelty search

    T2 - A study on game content generation

    AU - Liapis, Antonios

    AU - Yannakakis, Georgios N.

    AU - Togelius, Julian

    PY - 2015/3/16

    Y1 - 2015/3/16

    N2 - Novelty search is a recent algorithm geared toward exploring search spaces without regard to objectives. When the presence of constraints divides a search space into feasible space and infeasible space, interesting implications arise regarding hownovelty search explores such spaces. This paper elaborates on the problem of constrained novelty search and proposes two novelty search algorithms which search within both the feasible and the infeasible space. Inspired by the FI-2pop genetic algorithm, both algorithms maintain and evolve two separate populations, one with feasible and one with infeasible individuals, while each population can use its own selection method. The proposed algorithms are applied to the problem of generating diverse but playable game levels, which is representative of the larger problem of procedural game content generation. Results show that the two-population constrained novelty search methods can create, under certain conditions, larger and more diverse sets of feasible game levels than current methods of novelty search, whether constrained or unconstrained. However, the best algorithm is contingent on the particularities of the search space and the genetic operators used. Additionally, the proposed enhancement of offspring boosting is shown to enhance performance in all cases of two-population novelty search.

    AB - Novelty search is a recent algorithm geared toward exploring search spaces without regard to objectives. When the presence of constraints divides a search space into feasible space and infeasible space, interesting implications arise regarding hownovelty search explores such spaces. This paper elaborates on the problem of constrained novelty search and proposes two novelty search algorithms which search within both the feasible and the infeasible space. Inspired by the FI-2pop genetic algorithm, both algorithms maintain and evolve two separate populations, one with feasible and one with infeasible individuals, while each population can use its own selection method. The proposed algorithms are applied to the problem of generating diverse but playable game levels, which is representative of the larger problem of procedural game content generation. Results show that the two-population constrained novelty search methods can create, under certain conditions, larger and more diverse sets of feasible game levels than current methods of novelty search, whether constrained or unconstrained. However, the best algorithm is contingent on the particularities of the search space and the genetic operators used. Additionally, the proposed enhancement of offspring boosting is shown to enhance performance in all cases of two-population novelty search.

    UR - http://www.scopus.com/inward/record.url?scp=84924771106&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84924771106&partnerID=8YFLogxK

    U2 - 10.1162/EVCO_a_00123

    DO - 10.1162/EVCO_a_00123

    M3 - Article

    VL - 23

    SP - 101

    EP - 129

    JO - Evolutionary Computation

    JF - Evolutionary Computation

    SN - 1063-6560

    IS - 1

    ER -