Rule extraction from ensemble methods using aggregated decision trees

摘要: Ensemble methods have become very well known for being powerful pattern recognition algorithms capable of achieving high accuracy. However, produces learners that are not comprehensible or transferable thus making them unsuitable tasks require a rational justification decision. Rule Extraction can resolve this limitation by extracting rules from trained ensembles classifiers. In paper, we present an algorithm called REEMTIC uses symbolic learning (Decision Tree) on each underlying classifier the ensemble and combines them. Experiments theoretical analysis show generates highly accurate closely approximates Learned Model.