Date(s) - 08/10/2014
2:00 pm - 3:30 pm
We describe the conception of a visualization and transfor- mation tool for traces of the real-time strategy (RTS) computer game StarCraft. The development of our tool StarTrace is driven by the do- main the traces originate from as well as the observable elements those traces contain. We elaborate on those influences, which also include both the structure of the existing game traces and the requirement to use these traces to improve the performance of a machine learning (ML) agent that attempts to learn to play parts of the game. We then describe the architecture of the browser-based tool and the trace model behind it. The purpose of StarTrace is to eventually improve the learning process of this agent by providing the means to harness the enormous amount of data included in complex RTS games. Finally, an example application showcases how the tool can help to better understand the player behavior stored in game traces.
Speaker: Associate Professor Ian Watson, Department of Computer Science, The University of Auckland
Ian Watson has a PhD in Computer Science from Liverpool University. His career has involved the practical application of many areas of Artificial Intelligence research, including Knowledge Engineering and Expert Systems. His recent interests are in Case-Based Reasoning and Game AI.
For further information please contact Gregor White