Predicting Seminal Quality via Imbalanced Learning with Evolutionary Safe-Level Synthetic Minority Over-Sampling Technique

摘要: Seminal quality has fallen dramatically over the past two decades. Research indicates that environmental factors, health status, and life habits might lead to decline. Prediction of seminal is very useful in early diagnosis infertile patients. Recently, artificial intelligence (AI) technologies have been applied study male fertility potential. As it common many real applications about cognitive computation, prediction faces problem class imbalance, conventional algorithms are often biased towards majority class. In this paper, an evolutionary safe-level synthetic minority over-sampling technique (ESLSMOTE) proposed synthesize instances along same line with different weight degree, called safe level. The profile individual from dataset predicted via three classification methods ESLSMOTE. Important indicators, such as accuracy, precision, recall, receiver operating characteristic (ROC) curve, F1-score, used evaluate performance classifiers ESLSMOTE based on a tenfold cross-validation scheme. experimental results show can significantly improve accuracy back-propagation neural network, adaptive boosting, support vector machine. highest area under ROC curve (97.2%) given by ESLSMOTE-AdaBoost model. Experimental indicate ESLSMOTE-based outperform current state-of-the-art predicting terms curve. such, capability high accuracy.