Association Rule Mining as Knowledge Infusion into the Mechanics of an eXtended Classifier System

This article uses two well-established research areas: Numerical Association Rule Mining and the eXtended Classifier Systems. With their synergy, an attempt is made to advance the reuse of previously generated associated rules of a given dataset. Additionally, the article investigates how integrating infused rules into the XCS algorithm's population impacts its adaptive capabilities, and how well the learned knowledge transfers to testing dataset (environment) scenarios. This article also explores the novel approach of utilizing any user-selected dataset feature as a prioritized action for the eXtended Classifier System algorithm to adapt to. This adaptability is enabled by incorporating the prioritized action into the reinforcement learning process guided by an evaluation function. Such extended applicability should go beyond traditional classification tasks.