GMOTE: Gaussian based minority oversampling technique for imbalanced classification adapting tail probability of outliers

TL;DR

This paper introduces GMOTE, a novel oversampling technique for imbalanced classification that uses Gaussian Mixture Models and tail probability adaptation to generate more realistic synthetic minority instances and reduce outliers.

Contribution

The paper proposes GMOTE, a new oversampling method based on Gaussian Mixture Models and tail probability adaptation, improving over linear interpolation methods like SMOTE.

Findings

GMOTE outperforms SMOTE in accuracy and F1-score.

GMOTE effectively reduces outliers in synthetic data.

Experimental results show robust performance on benchmark datasets.

Abstract

Classification of imbalanced data is one of the common problems in the recent field of data mining. Imbalanced data substantially affects the performance of standard classification models. Data-level approaches mainly use the oversampling methods to solve the problem, such as synthetic minority oversampling Technique (SMOTE). However, since the methods such as SMOTE generate instances by linear interpolation, synthetic data space may look like a polygonal. Also, the oversampling methods generate outliers of the minority class. In this paper, we proposed Gaussian based minority oversampling technique (GMOTE) with a statistical perspective for imbalanced datasets. To avoid linear interpolation and to consider outliers, this proposed method generates instances by the Gaussian Mixture Model. Motivated by clustering-based multivariate Gaussian outlier score (CMGOS), we propose to adapt tail probability of instances through the Mahalanobis distance to consider local outliers. The experiment was carried out on a representative set of benchmark datasets. The performance of the GMOTE is compared with other methods such as SMOTE. When the GMOTE is combined with classification and regression tree (CART) or support vector machine (SVM), it shows better accuracy and F1-Score. Experimental results demonstrate the robust performance.