Autonomous classifiers with understandable rule using multi-objective genetic algorithms

This paper presents a method for designing autonomous classifiers via multi-objective genetic algorithms. The paper also proposes a novel objective measure to quantify the understandability of the classifiers. The other objectives of the classifiers are classification accuracy and average support value. We experimentally evaluate our approach on five different medical dataset and demonstrate that our algorithm encourages us to improve and apply this strategy in many real-world applications.