Albert Orriols Homepage

Bounding XCS’s Parameters for Unbalanced Datasets

by Albert Orriols-Puig and Ester Bernadó-Mansilla. Proceedings of the 8th annual Conference on Genetic and Evolutionary Computation. Pages 1561 - 1568. Seattle, WA.

This paper analyzes the behavior of the XCS classifier system on imbalanced datasets. We show that XCS with standard parameter settings is quite robust to considerable class imbalances. For high class imbalances, XCS suffers from biases toward the majority class. We analyze XCS’s behavior under such extreme imbalances and prove that appropriate parameter tuning improves significantly XCS’s performance. Specifically, we counterbalance the imbalance ratio by equalizing the reproduction probabilities of the most occurring and least occurring niches. The study provides guidelines to tune XCS’s parameters for unbalanced datasets, based on the dataset imbalance ratio. We propose a method to estimate the imbalance ratio during XCS’s training and adapt XCS’s parameters online.

No comments yet

Leave a Reply

You must be logged in to post a comment.