Bures-Wasserstein Barycenters and Low-Rank Matrix Recovery

Abstract

We revisit the problem of recovering a low-rank positive semidefinite matrix from rank-one projections using tools from optimal transport. More specifically, we show that a variational formulation of this problem is equivalent to computing a Wasserstein barycenter. In turn, this new perspective enables the development of new geometric first-order methods with strong convergence guarantees in Bures-Wasserstein distance. Experiments on simulated data demonstrate the advantages of our new methodology over existing methods.

Cite

Text

Maunu et al. "Bures-Wasserstein Barycenters and Low-Rank Matrix Recovery." Artificial Intelligence and Statistics, 2023.

Markdown

[Maunu et al. "Bures-Wasserstein Barycenters and Low-Rank Matrix Recovery." Artificial Intelligence and Statistics, 2023.](https://mlanthology.org/aistats/2023/maunu2023aistats-bureswasserstein/)

BibTeX

@inproceedings{maunu2023aistats-bureswasserstein,
  title     = {{Bures-Wasserstein Barycenters and Low-Rank Matrix Recovery}},
  author    = {Maunu, Tyler and Le Gouic, Thibaut and Rigollet, Philippe},
  booktitle = {Artificial Intelligence and Statistics},
  year      = {2023},
  pages     = {8183-8210},
  volume    = {206},
  url       = {https://mlanthology.org/aistats/2023/maunu2023aistats-bureswasserstein/}
}