An Accelerated Algorithm for Stochastic Bilevel Optimization under Unbounded Smoothness

This is code for NeurIPS 2024 paper "An Accelerated Algorithm for Stochastic Bilevel Optimization under Unbounded Smoothness"

Requirements

Pytorch 2.0, numpy, sklearn, tqdm

Run experiments

• To run AUC maximization, please refer to the directory auc_maximization.

• To run data hyper-cleaning, please refer to the directory data_cleaning.

Citation

If you found this repository helpful, please cite our paper:

@article{gong2024accelerated,
  title={An accelerated algorithm for stochastic bilevel optimization under unbounded smoothness},
  author={Gong, Xiaochuan and Hao, Jie and Liu, Mingrui},
  journal={The Thirty-eighth Annual Conference on Neural Information Processing Systems},
  year={2024}
}