AIConfPaper
ICML 2018oral31 citations

Frank-Wolfe with Subsampling Oracle

Thomas Kerdreux, Fabian Pedregosa, Alexandre d’Aspremont

Source ↗PDF ↗Scholar ↗TeX ↗

Abstract

We analyze two novel randomized variants of the Frank-Wolfe (FW) or conditional gradient algorithm. While classical FW algorithms require solving a linear minimization problem over the domain at each iteration, the proposed method only requires to solve a linear minimization problem over a small

BibTeX
@InProceedings{pmlr-v80-kerdreux18a,
  title = 	 {Frank-{W}olfe with Subsampling Oracle},
  author =       {Kerdreux, Thomas and Pedregosa, Fabian and d'Aspremont, Alexandre},
  booktitle = 	 {Proceedings of the 35th International Conference on Machine Learning},
  pages = 	 {2591--2600},
  year = 	 {2018},
  editor = 	 {Dy, Jennifer and Krause, Andreas},
  volume = 	 {80},
  series = 	 {Proceedings of Machine Learning Research},
  month = 	 {10--15 Jul},
  publisher =    {PMLR},
  pdf = 	 {http://proceedings.mlr.press/v80/kerdreux18a/kerdreux18a.pdf},
  url = 	 {https://proceedings.mlr.press/v80/kerdreux18a.html},
  abstract = 	 {We analyze two novel randomized variants of the Frank-Wolfe (FW) or conditional gradient algorithm. While classical FW algorithms require solving a linear minimization problem over the domain at each iteration, the proposed method only requires to solve a linear minimization problem over a small
Frank-Wolfe with Subsampling Oracle · ICML 2018