Path of Destruction: Learning an Iterative Level Generator Using a Small Dataset

Matthew Siper, Ahmed Khalifa, Julian Togelius

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

    Abstract

    We propose a new procedural content generation method which learns iterative level generators from a dataset of existing levels. The Path of Destruction method, as we call it, views level generation as repair; levels are created by iteratively repairing from a random starting level. The first step is to generate an artificial dataset from the original set of levels by introducing many different sequences of mutations to existing levels. In the generated dataset, features are observations of destroyed levels and targets are the specific actions that repair the mutated tile in the middle of the observations. Using this dataset, a convolutional network is trained to map from observations to their respective appropriate repair actions. The trained network is then used to iteratively produce levels from random starting maps. We demonstrate this method by applying it to generate unique and playable tile-based levels for several 2D games (Zelda, Danger Dave, and Sokoban) and vary key hyperparameters.

    Original languageEnglish (US)
    Title of host publicationProceedings of the 2022 IEEE Symposium Series on Computational Intelligence, SSCI 2022
    EditorsHisao Ishibuchi, Chee-Keong Kwoh, Ah-Hwee Tan, Dipti Srinivasan, Chunyan Miao, Anupam Trivedi, Keeley Crockett
    PublisherInstitute of Electrical and Electronics Engineers Inc.
    Pages337-343
    Number of pages7
    ISBN (Electronic)9781665487689
    DOIs
    StatePublished - 2022
    Event2022 IEEE Symposium Series on Computational Intelligence, SSCI 2022 - Singapore, Singapore
    Duration: Dec 4 2022Dec 7 2022

    Publication series

    NameProceedings of the 2022 IEEE Symposium Series on Computational Intelligence, SSCI 2022

    Conference

    Conference2022 IEEE Symposium Series on Computational Intelligence, SSCI 2022
    Country/TerritorySingapore
    CitySingapore
    Period12/4/2212/7/22

    Keywords

    • data augmentation
    • games
    • level generation
    • neural networks
    • procedural content generation
    • supervised learning

    ASJC Scopus subject areas

    • Artificial Intelligence
    • Computer Science Applications
    • Decision Sciences (miscellaneous)
    • Computational Mathematics
    • Control and Optimization
    • Transportation

    Fingerprint

    Dive into the research topics of 'Path of Destruction: Learning an Iterative Level Generator Using a Small Dataset'. Together they form a unique fingerprint.

    Cite this