Dissertation
Player modeling in adaptive games via multi-armed bandits
Doctor of Philosophy (Ph.D.), Drexel University
Dec 2022
DOI:
https://doi.org/10.17918/00001436
Abstract
Player models are an essential tool used by Experience Management AI when adapting a game experience for specific players. This research develops and explores approaches to player modeling in both single-player and multiplayer contexts that rely on the use of Multi-Armed Bandit (MAB) techniques. In addition to proposing advancements for MAB techniques to meet the challenges of this particular context, we extend our investigation into their use in more general and traditional applications, such as node selection policy in Monte Carlo Tree Search. We evaluate our approach toward both cooperative and competitive goals via simulated and real human user studies as well as adversarial board games.
Metrics
50 File views/ downloads
39 Record Views
Details
- Title
- Player modeling in adaptive games via multi-armed bandits
- Creators
- Robert C. Gray
- Contributors
- Santiago Ontañón (Advisor)Jichen Zhu (Advisor)
- Awarding Institution
- Drexel University
- Degree Awarded
- Doctor of Philosophy (Ph.D.)
- Publisher
- Drexel University; Philadelphia, Pennsylvania
- Number of pages
- xviii, 226 pages
- Resource Type
- Dissertation
- Language
- English
- Academic Unit
- Digital Media; Drexel University; Antoinette Westphal College of Media Arts and Design
- Other Identifier
- 991020041415004721