An imbalanced learning-based sampling method for physics-informed neural networks

Jiaqi Luo, Yahong Yang, Yuan Yuan, Shixin Xu, Wenrui Hao

Abstract: Highlights•RSmote: A novel local sampling method for PINNs using imbalanced learning.•Improves memory efficiency by targeting regions with large residuals.•Supported by theoretical analysis for reliable performance validation.•Outperforms RAD in accuracy and memory usage, especially in high dimensions.