Mario Plays on a Manifold: Generating Functional Content in Latent Space through Differential Geometry
This addresses the issue of ensuring functional content generation in games for developers and players, representing an incremental improvement over existing methods.
The paper tackles the problem of deep generative models creating non-functional content (like unsolvable game levels) by proposing a Riemannian geometry-based method for reliable interpolation and random walks in Categorical VAE latent spaces. Results show this geometry better maintains playable content in generated levels for 'Super Mario Bros' and 'The Legend of Zelda' compared to simpler baselines.
Deep generative models can automatically create content of diverse types. However, there are no guarantees that such content will satisfy the criteria necessary to present it to end-users and be functional, e.g. the generated levels could be unsolvable or incoherent. In this paper we study this problem from a geometric perspective, and provide a method for reliable interpolation and random walks in the latent spaces of Categorical VAEs based on Riemannian geometry. We test our method with "Super Mario Bros" and "The Legend of Zelda" levels, and against simpler baselines inspired by current practice. Results show that the geometry we propose is better able to interpolate and sample, reliably staying closer to parts of the latent space that decode to playable content.