AI RESEARCH
Data Balancing Strategies: A Systematic Survey of Resampling and Augmentation Methods
arXiv CS.AI
•
ArXi:2505.13518v2 Announce Type: replace-cross Imbalanced datasets, where one class significantly outnumbers others, remain a persistent challenge in machine learning, often biasing predictions toward the majority class and degrading classifier performance.