Fig. 9
From: An approach for handling imbalanced datasets using borderline shifting
Illustration of generating new minority instances using the k-nearest neighbors (k-NN) algorithm with Euclidean distance. For each minority instance, its k nearest minority neighbors are identified in the feature space, and synthetic samples are created along the line segments connecting the instance to its selected neighbors. This interpolation process enables the generation of new, realistic minority samples that enhance class balance while preserving the original data structure.
