arXiv paper of “Safe Screening for Unbalanced Optimal Transport“

Our new paper has been submitted to arXiv.

• Safe Screening for Unbalanced Optimal Transport

  • Author: Xun Su (M2, 2nd-year graduate student), Zhongxi Fang (D1), and HK
  • Abstract: This paper introduces a framework that utilizes the Safe Screening technique to accelerate the optimization process of the Unbalanced Optimal Transport (UOT) problem by proactively identifying and eliminating zero elements in the sparse solutions. We demonstrate the feasibility of applying Safe Screening to the UOT problem with ℓ2-penalty and KL-penalty by conducting an analysis of the solution’s bounds and considering the local strong convexity of the dual problem. Considering the specific structural characteristics of the UOT in comparison to general Lasso problems on the index matrix, we specifically propose a novel approximate projection, an elliptical safe region construction, and a two-hyperplane relaxation method. These enhancements significantly improve the screening efficiency for the UOT’s without altering the algorithm’s complexity.
  • Paper: arXiv preprint arXiv:2307.00247
  • Source code: GitHub.