70 likes | 250 Views
Dynamic Learning of AI in Gaming. Michael Hercenberg. Dynamic Learning. What is Dynamic Learning? Leaning in real-time Why? Replaces hardcoded scripts More ‘human’ game opponents More suitable environments Other aspects have advanced greatly Graphics Voicing Music Controls. Gaming AI.
E N D
Dynamic Learning of AI in Gaming Michael Hercenberg
Dynamic Learning • What is Dynamic Learning? • Leaning in real-time • Why? • Replaces hardcoded scripts • More ‘human’ game opponents • More suitable environments • Other aspects have advanced greatly • Graphics • Voicing • Music • Controls
Gaming AI • 80’s • Move left, then right • Wait for player • 90’s • Search-and-Destroy • Player-like Enemies • 00’s • Small advances • Loopholes
Methods • Data Mining • Uses data to create next level or encounter • Pro: Drastically alter the world • Con: Needs data and time
Methods • Difficulty Scaling • Makes changes in real-time • Pro: Makes corrections when needed • Con: Running real-time can be a strain
Methods • Explicit Context Tracking • ‘Windowed’ Data Mining
Future Research • Dynamic Choice Making • Introduce this into an RPG (Role-Playing Game) • Choices made drastically change the story • Not just give an alternate ending Choice Interactions Environment