Linear levels through n-grams

Steve Dahlskog, Julian Togelius, Mark J. Nelson

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Abstract

    We show that novel, linear game levels can be created using ngrams that have been trained on a corpus of existing levels. The method is fast and simple, and produces levels that are recognisably in the same style as those in the corpus that it has been trained on. We use Super Mario Bros. as an example domain, and use a selection of the levels from the original game as a training corpus. We treat Mario levels as a left-to-right sequence of vertical level slices, allowing us to perform level generation in a setting with some formal similarities to n-gram-based text generation and music generation. In empirical results, we investigate the effects of corpus size and n (sequence length). While the applicability of the method might seem limited to the relatively narrow domain of 2D games, we argue that many games in effect have linear levels and n-grams could be used to good effect, given that a suitable alphabet can be found. Copyright is held by the owner/author(s). Publication rights licensed to ACM.

    Original languageEnglish (US)
    Title of host publicationMINDTREK 2014 - Proceedings of the 18th International Academic MindTrek Conference: "Media Business, Management, Content and Services"
    PublisherAssociation for Computing Machinery, Inc
    Pages200-206
    Number of pages7
    ISBN (Print)9781450330060
    DOIs
    StatePublished - Nov 4 2014
    Event18th International Academic MindTrek Conference, MINDTREK 2014 - Tampere, Finland
    Duration: Nov 4 2014Nov 6 2014

    Other

    Other18th International Academic MindTrek Conference, MINDTREK 2014
    CountryFinland
    CityTampere
    Period11/4/1411/6/14

    Keywords

    • N-grams
    • Procedural content generation
    • Videogames

    ASJC Scopus subject areas

    • Computer Science Applications
    • Human-Computer Interaction
    • Software

    Cite this

    Dahlskog, S., Togelius, J., & Nelson, M. J. (2014). Linear levels through n-grams. In MINDTREK 2014 - Proceedings of the 18th International Academic MindTrek Conference: "Media Business, Management, Content and Services" (pp. 200-206). Association for Computing Machinery, Inc. https://doi.org/10.1145/2676467.2676506

    Linear levels through n-grams. / Dahlskog, Steve; Togelius, Julian; Nelson, Mark J.

    MINDTREK 2014 - Proceedings of the 18th International Academic MindTrek Conference: "Media Business, Management, Content and Services". Association for Computing Machinery, Inc, 2014. p. 200-206.

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Dahlskog, S, Togelius, J & Nelson, MJ 2014, Linear levels through n-grams. in MINDTREK 2014 - Proceedings of the 18th International Academic MindTrek Conference: "Media Business, Management, Content and Services". Association for Computing Machinery, Inc, pp. 200-206, 18th International Academic MindTrek Conference, MINDTREK 2014, Tampere, Finland, 11/4/14. https://doi.org/10.1145/2676467.2676506
    Dahlskog S, Togelius J, Nelson MJ. Linear levels through n-grams. In MINDTREK 2014 - Proceedings of the 18th International Academic MindTrek Conference: "Media Business, Management, Content and Services". Association for Computing Machinery, Inc. 2014. p. 200-206 https://doi.org/10.1145/2676467.2676506
    Dahlskog, Steve ; Togelius, Julian ; Nelson, Mark J. / Linear levels through n-grams. MINDTREK 2014 - Proceedings of the 18th International Academic MindTrek Conference: "Media Business, Management, Content and Services". Association for Computing Machinery, Inc, 2014. pp. 200-206
    @inproceedings{23b1d6111a14413eb01916e64bf5e778,
    title = "Linear levels through n-grams",
    abstract = "We show that novel, linear game levels can be created using ngrams that have been trained on a corpus of existing levels. The method is fast and simple, and produces levels that are recognisably in the same style as those in the corpus that it has been trained on. We use Super Mario Bros. as an example domain, and use a selection of the levels from the original game as a training corpus. We treat Mario levels as a left-to-right sequence of vertical level slices, allowing us to perform level generation in a setting with some formal similarities to n-gram-based text generation and music generation. In empirical results, we investigate the effects of corpus size and n (sequence length). While the applicability of the method might seem limited to the relatively narrow domain of 2D games, we argue that many games in effect have linear levels and n-grams could be used to good effect, given that a suitable alphabet can be found. Copyright is held by the owner/author(s). Publication rights licensed to ACM.",
    keywords = "N-grams, Procedural content generation, Videogames",
    author = "Steve Dahlskog and Julian Togelius and Nelson, {Mark J.}",
    year = "2014",
    month = "11",
    day = "4",
    doi = "10.1145/2676467.2676506",
    language = "English (US)",
    isbn = "9781450330060",
    pages = "200--206",
    booktitle = "MINDTREK 2014 - Proceedings of the 18th International Academic MindTrek Conference: {"}Media Business, Management, Content and Services{"}",
    publisher = "Association for Computing Machinery, Inc",

    }

    TY - GEN

    T1 - Linear levels through n-grams

    AU - Dahlskog, Steve

    AU - Togelius, Julian

    AU - Nelson, Mark J.

    PY - 2014/11/4

    Y1 - 2014/11/4

    N2 - We show that novel, linear game levels can be created using ngrams that have been trained on a corpus of existing levels. The method is fast and simple, and produces levels that are recognisably in the same style as those in the corpus that it has been trained on. We use Super Mario Bros. as an example domain, and use a selection of the levels from the original game as a training corpus. We treat Mario levels as a left-to-right sequence of vertical level slices, allowing us to perform level generation in a setting with some formal similarities to n-gram-based text generation and music generation. In empirical results, we investigate the effects of corpus size and n (sequence length). While the applicability of the method might seem limited to the relatively narrow domain of 2D games, we argue that many games in effect have linear levels and n-grams could be used to good effect, given that a suitable alphabet can be found. Copyright is held by the owner/author(s). Publication rights licensed to ACM.

    AB - We show that novel, linear game levels can be created using ngrams that have been trained on a corpus of existing levels. The method is fast and simple, and produces levels that are recognisably in the same style as those in the corpus that it has been trained on. We use Super Mario Bros. as an example domain, and use a selection of the levels from the original game as a training corpus. We treat Mario levels as a left-to-right sequence of vertical level slices, allowing us to perform level generation in a setting with some formal similarities to n-gram-based text generation and music generation. In empirical results, we investigate the effects of corpus size and n (sequence length). While the applicability of the method might seem limited to the relatively narrow domain of 2D games, we argue that many games in effect have linear levels and n-grams could be used to good effect, given that a suitable alphabet can be found. Copyright is held by the owner/author(s). Publication rights licensed to ACM.

    KW - N-grams

    KW - Procedural content generation

    KW - Videogames

    UR - http://www.scopus.com/inward/record.url?scp=84964056786&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84964056786&partnerID=8YFLogxK

    U2 - 10.1145/2676467.2676506

    DO - 10.1145/2676467.2676506

    M3 - Conference contribution

    SN - 9781450330060

    SP - 200

    EP - 206

    BT - MINDTREK 2014 - Proceedings of the 18th International Academic MindTrek Conference: "Media Business, Management, Content and Services"

    PB - Association for Computing Machinery, Inc

    ER -