Guided Rule Discovery in XCS for High-Dimensional Classification Problems

摘要: XCS is a learning classifier system that combines reinforcement scheme with evolutionary algorithms to evolve population of classifiers in the form condition-action rules. In this paper, we investigate effectiveness high-dimensional classification problems where number features greatly exceeds data instances --- common characteristics microarray gene expression tasks. We introduce new guided rule discovery mechanisms for XCS, inspired by feature selection techniques commonly used machine learning. The extracted quality information bias operators. performance proposed model compared standard and well-known using benchmark binary tasks sets. Experimental results suggests mechanism computationally efficient, promotes evolution more accurate solutions. performs significantly better than comparative when tackling problems.