A-SMOTE: A New Preprocessing Approach for Highly Imbalanced Datasets by Improving SMOTE

Abstract

Imbalance learning is a challenging task for most standard machine learning algorithms. The Synthetic Minority Oversampling Technique (SMOTE) is a well-known preprocessing approach for handling imbalanced datasets, where the minority class is oversampled by producing synthetic examples in feature vector rather than data space. However, many recent works have shown that the imbalanced ratio in itself is not a problem and deterioration of the model performance is caused by other reasons linked to the minority class sample distribution. The blind oversampling by SMOTE leads to two major problems: noise and borderline examples. Noisy examples are those from one class located in the safe zone of the other. Borderline examples are those located in the neighborhood of the class boundary. These samples are associated with deteriorating performance of the models developed. Therefore, it is critical to concentrate on the minority class data structure and regulate the positioning of the newly introduced minority class samples for better performance of classifiers. Hence, this paper proposes the advanced SMOTE, denoted as A-SMOTE, to adjust the newly introduced minority class examples based on distance to the original minority class samples. To achieve this objective, we first employ the SMOTE algorithm to introduce new samples to the minority and eliminate those examples that are closer to the majority than the minority. We apply the proposed method to 44 datasets at various imbalance ratios. Ten widely used data sampling methods selected from the literature are employed for performance comparison. The C4.5 and Naive Bayes classifiers are utilized for experimental validation. The results confirm the advantage of the proposed method over the other methods in almost all the datasets and illustrate its suitability for data preprocessing in classification tasks.

Effective treatment of imbalanced datasets in health care using modified SMOTE coupled with stacked deep learning algorithms

“…Usually in order not to avoid elimination of significant majority of instances, Oversampling algorithms are preferred, and Synthetic Minority Oversampling Technique (SMOTE) algorithm proposed by Chawla et al ( 2002 ) is the most widely used. Subsequently more than 85 variants of SMOTE have been reported in literature to further improve the basic form of SMOTE in terms of different classification metrics (Fernández et al 2018 ) like borderline-SMOTE1 and borderline-SMOTE2, advanced SMOTE (A-SMOTE), Distributed version of SMOTE (Han et al 2005 ; Hooda and Mann 2019 ; Hussein 2019 ), etc. There seems to be only few literature reports dealing with detailed critical comparison of these proposed methods (Bajer et al 2019 ; Kovács 2019 ).…”

Impacto de los algoritmos de sobremuestreo en la clasificación de subtipos principales del síndrome de guillain-barré

“…La suma de estos valores es de 40 y se divide por el número de filas de la tabla, para este caso es 18. El resultado es 2.222 y al ser el promedio más bajo, ocupa el número 1 en el ranking [30].…”

AdaBoost Ensemble Methods Using K-Fold Cross Validation for Survivability with the Early Detection of Heart Disease

“…Synthetic instances that are far from the borderline are easier to categorize than those that are near to the borderline, which present a significant learning difficulty for the majority of classifiers. The authors in [ 32 ] describe an advanced strategy (A-SMOTE) for preprocessing imbalanced training sets based on these findings. It aims to clearly characterize the borderline and create pure synthetic samples from SMOTE generalization.…”

“…AdaBoost makes it possible to merge various “weak classifiers” into a single classifier which is called “strong classifier.” Decision trees with one level, or decision trees with only one split, are the most popular algorithm used with AdaBoost. Decision Stump is another name for these trees [ 32 ]. This approach creates a model by assigning equal weights to all of the data points.…”