University of Twente Student Theses

Login

Generative Probabilistic Programming in Games: Creating Character Backgrounds Using a Bayesian Network

Li, Ruiyuan (2021) Generative Probabilistic Programming in Games: Creating Character Backgrounds Using a Bayesian Network.

[img] PDF
1MB
Abstract:Procedural content generation is a common facilitator in game design. Successful non-player characters (NPCs) backgrounds design can help with storytelling so that the players feel more involved and immersed in the game. This paper firstly discusses the definition of character background and how to select the related attributes based on archetypes in the template called monomyth (i.e., the hero's journey), then shows the way to formalize the characteristic design with the help of probabilistic grammars. Next, we realize the background generation via the Bayesian network and JS-divergence which serves as a metric for the performance evaluation. Finally, by updating conflicting rules and corresponding conditional probability tables in the Bayesian network, KL-divergence between the criteria and new observed distributions is minimized, which results in a local optimal realizing the improvement of the generator.
Item Type:Essay (Bachelor)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Computer Science BSc (56964)
Link to this item:https://purl.utwente.nl/essays/87325
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page