New QC-SMOTE method improves imbalanced classification accuracy

Researchers have developed QC-SMOTE, a novel oversampling framework designed to improve classification accuracy on imbalanced datasets. This method addresses the issue of generating low-quality synthetic samples by incorporating a quality-controlled approach that assesses sample reliability based on local density, safety levels, and isolation from the majority class. Experiments on 30 datasets demonstrate that QC-SMOTE outperforms existing oversampling techniques, achieving superior AUC-ROC and Macro F1 scores, particularly in scenarios with moderate to severe class imbalance. AI

IMPACT This research offers a more robust method for handling imbalanced datasets, potentially improving the performance of machine learning models in critical applications where data skew is common.

RANK_REASON The cluster describes a new academic paper detailing a novel algorithm for imbalanced classification.

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• AUC-ROC
• CatalyzeX Code Finder for Papers
• Connected Papers
• CORE Recommender
• DagsHub
• Gotit.pub
• Hugging Face
• IArxiv Recommender
• Influence Flower
• Litmaps
• Macro F1
• QC-SMOTE
• ScienceCast
• scite Smart Citations
• Smote

• paper
• other

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →