Local neighbourhood extension of SMOTE for mining imbalanced data

By Tomasz Maciejewski, Jerzy Stefanowski. 2011

256 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

Imbalanced Continual Learning with Partitioning Reservoir Sampling (2020, background): “More aggressively, there have also been directions that populate the minority samples via generation to avoid overfitting in the minority [14,58,23].”
Learning Deep Representation for Imbalanced Classification (2016, background): “To counter the negative effects, one often chooses from a few available options, which have been extensively studied in the past [7, 9, 11, 17, 18, 30, 40, 41, 46, 48].”
SMOTE for Learning from Imbalanced Data: Progress and Challenges, Marking the 15-year Anniversary (2018, background): “The interpolation mechanisms can be range restricted (Han et al., 2005; Bunkhumpornpat et al., 2009; Maciejewski & Stefanowski, 2011), for example by looking not only for nearest neighbours from the minority class but also from majority class; creating new examples closer to the selected instance…”
A Survey of Predictive Modeling on Imbalanced Domains (2016, background): “Nevertheless, SMOTE generates synthetic examples with the positive class label, disregarding the negative class examples that may lead to overgeneralization [Yen and Lee 2006; Maciejewski and Stefanowski 2011; Yen and Lee 2009].”
Class Rectification Hard Mining for Imbalanced Deep Learning (2017, background): “There are two classic approaches to learning from imbalanced data, (1) Class re-sampling : Either down-sampling the majority class or over-sampling the minority class or both [4, 11, 19, 20, 33, 36].”
Cross-Validation for Imbalanced Datasets: Avoiding Overoptimistic and Overfitting Approaches [Research Frontier] (2018, background): “However, for specific scenarios, SafeLevel-SMOTE may generate inconsistent examples [7]: if a minority example is an outlier, inside a well-defined majority cluster, then its slratio will be 0, causing the gap for SMOTE synthetization to be 1, thus creating a new minority instance in the exact locat”
SMOTE-IPF: Addressing the noisy and borderline examples problem in imbalanced classification by a re-sampling method with filtering (2015, methodology): “Within this group, the Safe-Levels-SMOTE (SL-SMOTE) [8], the Borderline-SMOTE (B1-SMOTE and B2-SMOTE) [23] or LN-SMOTE [37] methods are found, which try to create positive examples close to areas with a high concentration of positive examples or only inside the boundaries of the positive class.”
Dynamically Weighted Balanced Loss: Class Imbalanced Learning and Confidence Calibration of Deep Neural Networks (2021, methodology): “Data-level methods [16]–[23] alter the class distribution in the original data by employing resampling strategies to balance the data set.”
Cost-Sensitive Learning of Deep Feature Representations From Imbalanced Data (2015, background): “For example, Borderline SMOTE [16] only over-samples the minority class samples which lie close to the class boundaries.”
Deep Imbalanced Learning for Face Recognition and Attribute Prediction (2018, methodology): “mainly divided into two groups: data re-sampling [5], [6], [7], [9], [10], [11], [15] and cost-sensitive learning [8], [12], [13], [14], [16].”
Imbalanced Deep Learning by Minority Class Incremental Rectification (2018, methodology): “by down-sampling majority classes, over-sampling minority classes, or some combinations [5,6,7].”