Multi-stage ensemble with refinement for noisy labeled data classification

Chihyeon Choi, Woojin Lee, Youngdoo Son

Abstract: Highlights•We propose a novel multi-stage ensemble method for classification under noisy labels.•We combine several models capturing diverse features based on dataset refinement.•The dataset refinement does not require extra information on the corrupted dataset.•The proposed method outperforms existing methods on various noisy labeled datasets.