We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.OC

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Optimization and Control

Title: Sinkhorn Distributionally Robust Optimization

Abstract: We study distributionally robust optimization with Sinkorn distance -- a variant of Wasserstein distance based on entropic regularization. We derive convex programming dual reformulations when the nominal distribution is an empirical distribution and a general distribution, respectively. Compared with Wasserstein DRO, it is computationally tractable for a larger class of loss functions, and its worst-case distribution is more reasonable. To solve the dual reformulation, we propose an efficient batch gradient descent with a bisection search algorithm. Finally, we provide various numerical examples using both synthetic and real data to demonstrate its competitive performance.
Comments: 34 pages, 2 figures
Subjects: Optimization and Control (math.OC); Machine Learning (cs.LG); Machine Learning (stat.ML)
Cite as: arXiv:2109.11926 [math.OC]
  (or arXiv:2109.11926v1 [math.OC] for this version)

Submission history

From: Jie Wang [view email]
[v1] Fri, 24 Sep 2021 12:40:48 GMT (223kb,D)

Link back to: arXiv, form interface, contact.