Gaussian approximation for empirical barycenters
share
In this work we consider Wasserstein barycenters (average in Wasserstein distance) in Fourier basis. It includes a provement that a random Fourier parameter of the barycenter is close to a Gaussian random vector by distribution. The convergence rate is O(p/√(n)) depending on measures count (n) and the dimension of parameter (p).
READ FULL TEXT