Fig. 2
From: An approach for handling imbalanced datasets using borderline shifting
Illustration of the Borderline-Tomek process used to reduce class overlap. The figure shows how Tomek links are applied to identify pairs of nearest samples from opposite classesone minority and one majority instancebased on Euclidean distance. When a Tomek link is detected, the majority instance closest to the minority sample is removed, thereby cleaning the decision boundary and eliminating borderline noise. This step enhances class separability and prepares the data for subsequent oversampling33.
