Plenary Lecture

Plenary Lecture

A New Artificial Intelligence in Game Design - Introduction to Reinforcement Learning

Assistant Professor Shao-Shin Hung
Department of Computer Science and Information Engineering
WuFu Institute of Technology

Abstract: The past few years have seen steady improvements in computer technology for graphics, sound, networking and processing power. Computer-controlled, non-player characters facilitate games and activities in these worlds and may interact with hundreds of thousands of human-controlled characters. The game theory domain is been widely regarded as appropriate for understanding the concepts of machine learning. Scientists usually focus on strategic games and make efforts to create “intelligent” programs that efficiently compete with human players. Such games are suitable for further studying because of their complexity and the opportunities they offer to explore winning strategies However, artificial intelligence technology to control non-player characters has, so far, lagged behind advances in other virtual world technologies. There is now a need for more believable and intelligent non-player characters to support and enhance virtual world applications.
This speech presents a new artificial intelligence technique – motivated reinforcement learning – for the development of non-player characters in multiuser games. On the other side, reinforcement learning is considered as one of the most suitable and prominent methods for solving game problems due to its capability to discover good strategies by extended self-training and limited initial knowledge. For example, humans and animals have the ability to focus and adapt their behavior. These behavioral traits are also an advantage for artificial agents in complex or dynamic environments, where only a small amount of available information may be relevant at a particular time, and relevant information changes over time. Motivated reinforcement learning combines computational models of motivation with advanced machine learning algorithms – to empower non-player characters to self identify new tasks on which to focus their attention and learn about.
Finally, both theoretical and practical issues are addressed for developing adaptive, dynamic non-player characters. Focus applications include multiuser, role-playing and simulation games.

Brief Biography of the Speaker:
Shao-Shin Hung received the MS. and Ph.D. degrees Computer Science and Information Engineering from National Cheng Chung University Taiwan, in 1992 and 2007, respectively.
Currently, he is an Assistant Professor at the Department of Computer Science and Information Engineering, WuFu Institute of Technology. He serves as a program committee of the 2nd Int. Multi-Conference on Engineering and Technological Innovation (IMETI09), the 4th International Conference on Ubiquitous Information Technologies & Applications (ICUT 2009). He also serves as an Associate Editor/Editorial Board member of the following international journals, such as the Open Software Engineering Journal, the Open Industrial and Manufacturing Engineering Journal, and the Open Artificial Intelligence Journal. He is a paper reviewer of Vis'07, Vis'08, VAST'08, Vis'09 and Journal of Information Science.
His research interests include computational intelligence, data mining, intrusion detection and applications of 3D game system tools. He is a member of the ACM and the IEEE Computer Society.

WSEAS Unifying the Science