The KNN-Equality [1] selects an equal number of examples for each class. It is a good alternative to define the region of competence for imbalanced datasets [2].
[2] Cruz, Rafael MO, Dayvid VR Oliveira, George DC Cavalcanti, Robert Sabourin. "FIRE-DES++: Enhanced online pruning of base for dynamic ensemble selection." Recognition 85 (2019): 149-160.
Menelau
commented
4 years ago
The KNN-Equality [1] selects an equal number of examples for each class. It is a good alternative to define the region of competence for imbalanced datasets [2].
Refs: [1] Sierra, Basilio, Elena Lazkano, Itziar Irigoien, Ekaitz Jauregi, Iñigo Mendialdua. "K nearest neighbor equality: giving equal chance all existing classes."Sciences 181, no. 23 (2011): 5158-
[2] Cruz, Rafael MO, Dayvid VR Oliveira, George DC Cavalcanti, Robert Sabourin. "FIRE-DES++: Enhanced online pruning of base for dynamic ensemble selection." Recognition 85 (2019): 149-160.