Tag
This paper challenges the standard prescription of balancing datasets to avoid spurious correlations, showing that in a synthetic sum parity task with two-layer transformers, high data imbalance (spurious ratio 0.9) promotes robust generalization while low imbalance (0.5) hinders it, through a mechanism of shortcut saturation.
Proposes FedEAS, a budget-aware policy for synthetic data augmentation in federated learning that assigns each client an entropy-adaptive per-class generation budget, recovering most accuracy gains of full class balancing while reducing generation cost by 94.1%.
This paper systematically evaluates five imbalance handling methods (RUS, ROS, SMOTE, re-weighting, direct F1 optimization) on three biomedical datasets (tabular, text, image) using models of varying complexity. Results show that benefits depend on model complexity and data modality, with ROS, re-weighting, and direct F1 optimization being effective for complex models on unstructured data.