Bringing Regularized Optimal Transport to Lightspeed: A Splitting Method Adapted for GPUs

Abstract

We present an efficient algorithm for regularized optimal transport. In contrast toprevious methods, we use the Douglas-Rachford splitting technique to developan efficient solver that can handle a broad class of regularizers. The algorithmhas strong global convergence guarantees, low per-iteration cost, and can exploitGPU parallelization, making it considerably faster than the state-of-the-art formany problems. We illustrate its competitiveness in several applications, includingdomain adaptation and learning of generative models.

PDF NeurIPS OpenReview Semantic Scholar

Cite

Text

Lindbäck et al. "Bringing Regularized Optimal Transport to Lightspeed: A Splitting Method Adapted for GPUs." Neural Information Processing Systems, 2023.

Markdown

[Lindbäck et al. "Bringing Regularized Optimal Transport to Lightspeed: A Splitting Method Adapted for GPUs." Neural Information Processing Systems, 2023.](https://mlanthology.org/neurips/2023/lindback2023neurips-bringing/)

BibTeX

@inproceedings{lindback2023neurips-bringing,
  title     = {{Bringing Regularized Optimal Transport to Lightspeed: A Splitting Method Adapted for GPUs}},
  author    = {Lindbäck, Jacob and Wang, Zesen and Johansson, Mikael},
  booktitle = {Neural Information Processing Systems},
  year      = {2023},
  url       = {https://mlanthology.org/neurips/2023/lindback2023neurips-bringing/}
}