Class-specific intuitionistic fuzzy kernel ridge regression classifier

In real-world data classification problems, class imbalance learning, noise and outliers are the major problems. The conventional kernel ridge regression (KRR) cannot efficiently deal with these challenges because all the samples are provided equal importance irrespective of their contribution to decision-making. Hence, to address this problem, we suggest a novel class-specific intuitionistic fuzzy KRR (CS-IFKRR) model for classification. CS-IFKRR provides appropriate weights to the samples for effective decision-making. CS-IFKRR classifier is designed to tackle the challenge of class imbalance in classification tasks, which generally leads to biased predictions and poor generalization for minority classes. Moreover, time efficiency is a secondary but significant advantage of CS-IFKRR, as it solves systems of linear equations. In addition to that intuitionistic fuzzy score values consider sample distance and heterogeneity to determine appropriate weights. The experimental investigation is carried out over a few popular datasets. The classification performance of the proposed CS-IFKRR model is contrasted with that of support vector machine (SVM), twin SVM, intuitionistic fuzzy SVM (IFSVM), IF twin SVM (IFTSVM), KRR and intuitionistic fuzzy KRR (IFKRR). The results, based on accuracy, F1 score and G-mean reveal the superiority of CS-IFKRR over other relevant models. Further statistical analysis is carried out based on Friedman test and posthoc Nemenyi analysis.