Human in the Loop Fuzzy Pattern Tree Evolution

Abstract Fuzzy pattern trees evolved using grammatical evolution, a system we call Grammatical Evolution, are shown to be robust Explainable Artificial Intelligence technique. Experimental results show Evolution achieves competitive when compared against SVM, Random Forest and Logistic Regression on set of real world benchmark problems. allows for human input throughout the evolutionary process. Regularization methods double tournament selection were investigated determine what method was most successful at finding smaller, more interpretable models. A domain expert recruited investigate interpretability models found give confidence score each model. This successfully identified overfit models, validating that can regarded powerful