Conference proceeding
Procedural level generation using multi-layer level representations with MdMCs
2017 IEEE Conference on Computational Intelligence and Games (CIG)
Aug 2017
Abstract
The use of statistical and machine learning approaches, such as Markov chains, for procedural content generation (PCG) has been growing in recent years in the field of Game AI. However, many of these level generation approaches account for only the structural properties of the levels. We developed multi-layered representations of levels, where each layer is designed to capture distinct gameplay information. Specifically, we experiment with representing levels using three layers: a structural layer that captures the object placements in the level, a player path layer that captures the path of an agent through the level, and a height layer that captures sections of the level as various height groups. We test our approach by generating levels for Super Mario Bros. with a multi-dimensional Markov chain approach. We compare the levels sampled with our multi-layered representation with those sampled using a single- layer approach that only considers the structural layer.
Metrics
11 Record Views
13 citations in Scopus
Details
- Title
- Procedural level generation using multi-layer level representations with MdMCs
- Creators
- Sam Snodgrass - Drexel UniversitySantiago Ontanon - Drexel University
- Publication Details
- 2017 IEEE Conference on Computational Intelligence and Games (CIG)
- Conference
- 2017 IEEE Conference on Computational Intelligence and Games (CIG)
- Publisher
- IEEE
- Resource Type
- Conference proceeding
- Language
- English
- Academic Unit
- Computer Science
- Scopus ID
- 2-s2.0-85040024238
- Other Identifier
- 991019173756604721