CLE-SMOTE: Addressing Extreme Imbalanced Data Classification with Contrastive Learning-Enhanced SMOTE
Keywords: Class Imbalance, Data Augmentation, Deep Learning, Contrastive Learning, SMOTE, Noisy Data
Abstract: Synthetic Minority Oversampling Technique (SMOTE) is a widely used oversampling method for addressing class imbalance by generating synthetic minority class examples. While effective, SMOTE occasionally introduces harmful examples into the dataset, hindering model
performance. In this work, we introduce Contrastive Learning-Enhanced SMOTE (CLESMOTE), a method to identify and reduce the influence of these noisy SMOTE-generated examples. In our experiments on imbalanced datasets, CLE-SMOTE achieves promising results, substantially outperforming all baselines, including vanilla SMOTE, and approaching the performance of an equivalent network trained on a balanced dataset.
Primary Subject Area: Domain specific data issues
Paper Type: Extended abstracts: up to 2 pages
DMLR For Good Track: Participate in DMLR for Good Track
Participation Mode: In-person
Confirmation: I have read and agree with the workshop's policy on behalf of myself and my co-authors.
Submission Number: 81
Loading