Page 301 - Special Topic Session (STS) - Volume 1
P. 301
STS430 Eustasio D.B. et al.
⁄
⁄
′
[ℒ [{ , ()} 1 ] , ℒ [{ ′ , ()} 1 ]] ≤ 1 ∑ ( , ).
=1
Hence, the Wasserstein distance of the variance of two collections of
distributions can be controlled using the distance between the distributions.
The main consequence of this fact is that the minimal alignment cost can also
be bootstrapped as soon as a distributional limit theorem exists for , (), as
in the discussion above. In Sections ?? and ?? below, we present distributional
results of this type in the one-dimensional case. We note that, while general
Central Limit Theorems for the empirical transportation cost are not available
in dimension > 1, some recent progress has been made in this direction;
see, e.g., [30] for Gaussian distributions and [32], which gives such results for
distributions on ℝ with finite support. Further advances along these lines
would make it possible to extend the results in the following section to higher
dimensions.
References
1. M. Agueh, G. Carlier, Barycenters in the Wasserstein space, SIAM J. Math.
Anal. 43 (2011) 904–924.
2. M.Agulló-Antolín, J.A.Cuesta-Albertos, H.Lescornel, J.-M.Loubes, A
parametric registration model for warped distributions with
Wasserstein’s distance, J. Multivariate Anal. 135 (2015) 117–130.
3. S. Allassonnière, Y. Amit, A. rouvé, Towards a coherent statistical
framework for dense deformable template estimation, J. R. Stat. Soc. Ser.
B (Stat. Methodol.) 69 (2007) 3–29.
4. P.C. Álvarez-Esteban, E. del Barrio, J.A. Cuesta-Albertos, C. Matrán,
Trimmed comparison of distributions, J. Amer. Statist. Assoc. 103 (2008)
697–704.
5. Y. Amit, U. Grenander, M. Piccioni, Structural image restoration through
deformable template, J. Amer. Statist. Assoc. 86 (1991) 376–387.
6. J. Bigot, T. Klein, Characterization of barycenters in the Wasserstein
space by averaging optimal transport maps, ESAIM: Probability and
Statistics, in press (2018).
7. S. Bobkov, M. Ledoux, One-dimensional empirical measures, order
statistics and Kantorovich transport distances, Memoirs of the American
Mathematical Society, in press (2018).
8. E. Boissard, T. Le Gouic, J.-M. Loubes, Distribution’s template estimate
with Wasserstein metrics, Bernoulli 21 (2015) 740–759.
9. DelBarrio,EustasioandGordaliza,PaulaandLescornel,Hélèneand Loubes,
Jean-Michel, Central limit theorem and bootstrap procedure for
Wassersteinâs variations with an application to structural relationships
between distributions, Journal of Multivariate Analysis,169,341–362.
290 | I S I W S C 2 0 1 9