Modeling XCS in class imbalances: Population sizing and parameter settings

The document discusses the challenges and strategies in modeling XCS (eXtended Classifier System) to handle class imbalances in datasets, particularly focusing on population size and parameter settings. It analyzes the ability of XCS to create effective rules for minority classes and investigates performance in various testing scenarios, including the one-bit problem. The findings emphasize the importance of appropriate population size and learning parameters to ensure proper representation of starved niches in imbalanced classification tasks.