Loading...
Please wait, while we are loading the content...
Similar Documents
Generalization in evolutionary learning (1997).
| Content Provider | CiteSeerX |
|---|---|
| Author | Wilson, S. W. |
| Abstract | In artificial life, evolutionary techniques are widely employed for adaptation over generations, but not for learning. Holland's classifier system concept proposed that learning could also be an evolutionary process. Traditional classifier systems have not worked well, but a recently developed, non-standard system, XCS (Wilson 1995), showed substantially improved performance and---essential for artificial creatures---a powerful generalization ability. In XCS, classifier fitness is based on prediction accuracy and the genetic algorithm takes place in environmental niches. The present paper reports on two changes to XCS that were aimed at increasing its tendency to evolve accurate, maximally general classifiers and were tested on previously employed "woods" and multiplexer tasks. Together the changes bring XCS close to evolving populations whose high-fitness classifiers form a near-minimal, accurate, maximally general cover of the input and action product space. In addition, results on t... |
| File Format | |
| Publisher Date | 1997-01-01 |
| Access Restriction | Open |
| Subject Keyword | Evolutionary Learning Artificial Creature Prediction Accuracy Non-standard System High-fitness Classifier Genetic Algorithm Evolutionary Process Evolutionary Technique General Cover General Classifier Action Product Space Multiplexer Task Powerful Generalization Ability Artificial Life Classifier System Concept Present Paper Report Traditional Classifier System Environmental Niche |
| Content Type | Text |
