ML internals: Synthetic Minority Oversampling (SMOTE) Technique
Domino Data Lab
MAY 20, 2021
propose a different strategy where the minority class is over-sampled by generating synthetic examples. Their tests are performed using C4.5-generated note that this variant “performs worse than plain under-sampling based on AUC” when tested on the Adult dataset (Dua & Graff, 2017). In their 2002 paper Chawla et al.
136 
Let's personalize your content