Active fuzzy rule induction

The use of rule based learners has been highly motivated all these years because of their inherent properties of interpretability and comprehensibility, leading to the construction of user friendly exported models by keeping pace with propositional logic. Besides this, their ability to operate under efficient time complexity allows us to occupy it under Active Learning schemes that integrate the human factor as an oracle into their learning kernel so as to tackle with the scarcity of existing labeled examples over several scientific fields. Upon this assumption, a recently proposed fuzzy rule based learner has been combined with a suitable query strategy for mining, with both robust and fast enough ability, unlabeled instances that facilitate the improvement of the learning behavior of the whole classification method. Rigorous experiments have been executed, proving the rightness of our ambition.