VFC-SMOTE: very fast continuous synthetic minority oversampling for evolving data streams

Abstract

The world is constantly changing, and so are the massive amount of data produced. However, only a few studies deal with online class imbalance learning that combines the challenges of class-imbalanced data streams and concept drift. In this paper, we propose the very fast continuous synthetic minority oversampling technique (VFC-SMOTE). It is a novel meta-strategy to be prepended to any streaming machine learning classification algorithm aiming at oversampling the minority class using a new version of Smote and Borderline-Smote inspired by Data Sketching. We benchmarked VFC-SMOTE pipelines on synthetic and real data streams containing different concept drifts, imbalance levels, and class distributions. We bring statistical evidence that VFC-SMOTE pipelines learn models whose minority class performances are better than state-of-the-art. Moreover, we analyze the time/memory consumption and the concept drift recovery speed.

Class balancing framework for credit card fraud detection based on clustering and similarity-based selection (SBS)

“…To deal with this point, there are a lot of researchers worked to extend the original version of SMOTE. For instance, SMOTE-TL [ 17 ], SMOTE-ENN [ 28 ], Borderline SMOTE [ 29 ], MSMOTE [ 30 ], VFC-SMOTE [ 31 ] and etc.…”

A survey on learning from imbalanced data streams: taxonomy, challenges, empirical study, and reproducible experimental framework

“…They are calculated over a sliding window of 500 instances. (Bernardo et al, 2020a) Data-Level RI Online Explicit C-SMOTE (Bernardo et al, 2020b) Data-Level RI Online Explicit VFC-SMOTE (Bernardo and Della Valle, 2021b) Data-Level RI Online Explicit CSARF (Loezer et al, 2020) Algo-Level CS Online Explicit -GHVFDT (Lyon et al, 2014) Algo-Level TM Online Implicit -HDVFDT (Cieslak and Chawla, 2008) Algo-Level TM Online Implicit -ARF (Gomes et al, 2017) Ensemble -Online Explicit -KUE (Cano and Krawczyk, 2020) Ensemble -Hybrid Explicit LB (Bifet et al, 2010a) Ensemble -Online Explicit OBA (Bifet et al, 2009) Ensemble -Online Explicit SRP (Gomes et al, 2019) Ensemble -Online Explicit ESOS-ELM (Mirza et al, 2015) Ensemble -Chunk Explicit -CALMID (Liu et al, 2021) Ensemble -Hybrid Explicit MICFOAL (Liu et al, 2021) Ensemble -Online Explicit ROSE Ensemble -Hybrid Explicit OADA (Wang and Pineau, 2016) Ensemble -Online Explicit OADAC2 (Wang and Pineau, 2016) Ensemble -Online Explicit ARFR (Ferreira et al, 2019) Ensemble RI Online Explicit -SMOTE-OB (Bernardo and Della Valle, 2021a) Ensemble RI Online Explicit OSMOTE (Wang and Pineau, 2016) Ensemble RI Online Explicit OOB Ensemble ROB Online Implicit UOB Ensemble RUB Online Implicit ORUB (Wang and Pineau, 2016) Ensemble RUB Online Explicit OUOB (Wang and Pineau, 2016) Ensemble RB Online Explicit Kappa is used to evaluate classifiers in imbalanced settings (Brzeziński et al, , 2019. It evaluates the classifier performance by computing the inter-rater agreement between the successful predictions and the statistical distribution of the data classes, correcting agreements that occur by mere statistical chance.…”