Loading...
Please wait, while we are loading the content...
Similar Documents
Evolution of Reinforcement Learning in Games or How to Win against Humans with Intelligent Agents
| Content Provider | Semantic Scholar |
|---|---|
| Author | Pignede, Thomas |
| Copyright Year | 2012 |
| Abstract | This paper reviews reinforcement learning in games. Games are a popular domain for the design and evaluation of AI systems, and still offer formidable challenges. This paper presents some classic and more recent work from this field. First it starts with Tesauro’s TD-Gammon, which was one of the first successes where a game-playing agent learned only from its own experience. Then it talks about Hu and Wellman’s approach for modelling multi-agent environments and calculating their Nash-equilibria. Finally it explains Johanson, Zinkevich and Bowling’s method on how to deal with stochasticity and computing robust strategies. The conclusion connects the different algorithms together and gives an outlook on current research topics and on the practical application of the presented techniques. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://www.ias.informatik.tu-darmstadt.de/uploads/Teaching/AutonomousLearningSystems/Pignede_ALS_2012.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
