Synthetic Minority Over-Sampling Technique (SMOTE) is a data-augmentation approach for classification problems which introduces synthetic examples from low-representation classes by interpolating features from neighboring samples.