Corpus ID: 219966115

On Projection Robust Optimal Transport: Sample Complexity and Model Misspecification

@article{Lin2020OnPR,
  title={On Projection Robust Optimal Transport: Sample Complexity and Model Misspecification},
  author={Tianyi Lin and Zeyu Zheng and Elynn Y. Chen and Marco Cuturi and Michael I. Jordan},
  journal={ArXiv},
  year={2020},
  volume={abs/2006.12301}
}
  • Tianyi Lin, Zeyu Zheng, +2 authors Michael I. Jordan
  • Published 2020
  • Mathematics, Computer Science
  • ArXiv
  • Optimal transport (OT) distances are increasingly used as loss functions for statistical inference, notably in the learning of generative models or supervised learning. Yet, the behavior of minimum Wasserstein estimators is poorly understood, notably in high-dimensional regimes or under model misspecification. In this work we adopt the viewpoint of projection robust (PR) OT, which seeks to maximize the OT cost between two measures by choosing a k-dimensional subspace onto which they can be… CONTINUE READING

    Figures from this paper.

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 77 REFERENCES
    Adam: A Method for Stochastic Optimization
    • 49,659
    • Highly Influential
    • PDF
    Improved Training of Wasserstein GANs
    • 3,332
    • PDF
    Convergence of probability measures
    • 3,522
    • PDF
    Sinkhorn Distances: Lightspeed Computation of Optimal Transport
    • 980
    • PDF
    Optimal Transport: Old and New
    • 3,409
    • Highly Influential
    Wasserstein Generative Adversarial Networks
    • 1,969
    • Highly Influential
    • PDF
    Fast Computation of Wasserstein Barycenters
    • 351
    • PDF
    Infinite Dimensional Analysis: A Hitchhiker’s Guide
    • 1,699
    • PDF
    Wasserstein Auto-Encoders
    • 331
    • PDF
    Sliced Wasserstein Kernel for Persistence Diagrams
    • 97
    • PDF