Combining Supervised and Unsupervised Fuzzy Learning Algorithms for Robust Diabetes Diagnosis

In domains that have complex data characteristics and/or noisy data, any single supervised learning algorithm tends to suffer from overfitting. One way mitigate this problem is combine unsupervised component as a front end of the main learner. paper, we propose hierarchical combination fuzzy C-means clustering and max–min neural network learner for purpose. The proposed method evaluated in domain (Pima Indian Diabetes open database). showed superior result standalone backpropagation-based network. also better performance than tested same literature with high accuracy (80.96%) was at least competitive other measures such sensitivity, specificity, F1 measure.