Loading...
Please wait, while we are loading the content...
Similar Documents
Advanced strategy representations for the iterated prisoner ’ s dilemma
| Content Provider | Semantic Scholar |
|---|---|
| Author | Slodzinski, Rafael Ehlers, Rüdiger Seis, Danny Bartz-Beielstein, Thomas Schmitt, Karlheinz Mehnen, Jörn |
| Copyright Year | 2015 |
| Abstract | Although many studies have been conducted on evolving strategies for the iterated prisoners dilemma (IPD) using evolutionary algorithms, most of them use rather simple representations like look-up tables. Knowledge about the evolution of strategies with complex representations remains limited. Complex representations may lead to new strategies and may be closer to the human decision making process (“thinking"’). Our main question is: Do complex represantations have substantial advantages over simpler ones in context of the IPD? In this paper we introduce two novel advanced representations: an exhaustive approach based on linear genetic programming and an evolution strategy based approach which reflects a psychological model. We show that both approaches lead to the generation of usable tournament strategies. We also use a cooevolutionary environment to study the evolution of cooperation, leading to diverse results. Additionally, the application of decision tree techniques to generate a strategy training set from real world tournaments is discussed. Strategies evolved with such training sets may perform well in subsequent real world tournaments, as other participants tend to imitate successful strategies of previuos tournaments. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://www.spotseven.de/wp-content/papercite-data/pdf/bara06a.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
