This repo contains the code for NeurIPS 2021 paper:

• Heavy Ball Momentum for Conditional Gradient

• prob.py defines the loss functions and constraint sets.
• optimizer.py implements proposed and benchmark algorithms.
• main.ipynb gives an example of binary classification on dataset mushroom.

If the code is helpful for your research, please cite the following paper

@inproceedings{li2021heavy,
  title={Heavy Ball Momentum for Conditional Gradient},
  author={Li, Bingcong and Sadeghi, Alireza and Giannakis, Georgios B},
  booktitle={Proc. Advances in Neural Info. Process. Syst.},
  year={2021}
}