PhD Course - Advanced Topics in Procedural Content Generation
Sebastian Risi, Associate Professor
Martin Pichlmair, Associate Professor
Niels Justesen, PhD Fellow
Mads Johansen Lassen, PhD Fellow
Georgios N. Yannakakis, Professor
Michael Cook, Senior Research Fellow
Gabriella Alves Bulhões Barros, PhD Student
Joris Dormans, PhD Student
Date(s) of the course: 5. august – 9. August
Location and Format: Lectures and project presentations will be streamed online. Students can thus participate across the globe from their own institution but are also welcome to show up at the ITU and NYU. A slack group will be set up to handle communication during and in-between the lectures.
Time: Lectures will be scheduled from Monday to Wednesday in the afternoon (CEST) and presentations at Friday afternoon. The rest of the week should be spent on the mini-project.
Course description: Procedural Content Generation (PCG) is the algorithmic creation of content within video games, such as levels, rules, mechanics, textures, stories, music, characters, etc. PCG can for example be used to adapt a game as players are interacting with it or as a tool to assist game designers in the
This Ph.D. course will consist of a series of lectures that will present the participants with advanced topics in PCG from leading researchers in the field and prominent practitioners in the game industry. This course is aimed for Ph.D. student’s already working on procedural systems or those who wish to apply procedural algorithms in their project. The course consists of three activities: 1) reading the material prior to the course, 2) “attending” to lectures, 3) carry out a mini-project based on material from the lecturers.
Recent developments in machine learning is opening up for new possibilities for PCG. However, modern machine learning algorithms can be difficult to analyze and understand. The topic of this course will thus be focused around machine learning for PCG, how to analyze the output of a PCG system, and how to incorporate human evaluation or guidance in the generation procedure.
When students have completed the course, they will be able to:
- describe, explain and implement basic and advanced PCG algorithms
- analyze and explain the output of a PCG system
- account for and reflect on how PCG can support game development/design
Programme: The tentative programme of the course in Central European Summer Time (CEST). Participants are responsible for managing their own time to work on their mini-project during the week.
Monday, August 5th:
|Martin Pichlmair / Sebastian Risi ||15.00 - 15.15 ||Welcome and introduction to the course |
|Lecturer 1 ||15.15 - 16.45 || |
|Lecturer 2 ||17.00 - 18.00 || |
|Lecturer 3 ||18.15 - 19.15 || |
Tuesday, August 6th:
|Lecturer 1 ||15.00 - 16.30 || |
|Lecturer 2 ||16.45 - 17.45 || |
|Lecturer 3 ||18.00 - 19.00 || |
Wednesday, August 7th:
|Lecturer 4 ||15.15 - 16.46 || |
|Lecturer 5 ||17.00 - 18.00 || |
Thursday, August 8th:
|No lectures - only project work |
Friday, August 9th:
|Presentations ||15 - ||Max. 15 minutes per participant |
Mini-project: Each participant is expected to work on a project related to the topics covered in the course. The project can e.g. be a digital prototype of generative system or an analysis of an existing system.
Prerequisites: Basic programming skills are required. Additionally, knowledge on some of the following areas is preferred: algorithms, design, machine learning and optimization.
Exam: The participants will be evaluated on the mini-project carried out during the course that will be presented on the last day of the course. The course organizers will ask question and provide feedback on both the technical and theoretical part of the project. Participants will receive a signed
diploma for their participation.
Credits: 3 ECTS
Amount of hours the student is expected to use on the course:
Participation: 38 hours: 11 hours of lectures, 24 hours of project work, 3 hours of presentations.
Preparation: 42 hours of reading. We expect the readings to be around 250 pages.
Total: 80 hours
How to sign up: Niels Justesen, firstname.lastname@example.org
- Chapters 1, 2, 3, 8, 9, 10, 11, 12 of Procedural Content Generation in Games - A textbook and an overview of current research (Noor Shaker, Julian Togelius, and Mark J. Nelson). http://pcgbook.com/
- Summerville, Adam, et al. "Procedural content generation via machine learning (PCGML)." IEEE Transactions on Games 10.3 (2018): 257-270. https://arxiv.org/pdf/1702.00539.pdf
- Zhu, Jichen, et al. "Explainable AI for Designers: A Human-Centered Perspective on Mixed-Initiative Co-Creation" http://sebastianrisi.com/wp-content/uploads/zhu_cig18.pdf
- Novick, David G. and Sutton, Stephen “What is Mixed-Initiative Interaction? https://pdfs.semanticscholar.org/f4e7/2c1f3834307095c8d71f6e44341106d12d45.pdf
- Yannakakis, Georgios et al. “Mixed-Initaitive Co-Creativity” http://www.antoniosliapis.com/papers/mixed_initiative_co-creativity.pdf
- The lecturers will propose additional papers and/or book chapters.