Extending Bagging for Imbalanced Data
[ 1 ] Instytut Informatyki, Wydział Informatyki, Politechnika Poznańska | [ P ] pracownik
2013
referat
angielski
EN Various modifications of bagging for class imbalanced data are discussed. An experimental comparison of known bagging modifications shows that integrating with undersampling is more powerful than oversampling. We introduce Local-and-Over-All Balanced bagging where probability of sampling an example is tuned according to the class distribution inside its neighbourhood. Experiments indicate that this proposal is competitive to best undersampling bagging extensions.
269 - 278
WoS (15)