Skip to content

Convolutional Wasserstein Distances, in Python, based on a lot of peoples work.

Notifications You must be signed in to change notification settings

gionuno/conv_wasserstein_dist

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

conv_wasserstein_dist

Convolutional Wasserstein Distances, in Python, based on a lot of peoples work.

image

So, basically this uses, a Wasserstein distance to calculate barycenters of distributions. The idea being that the correct method for finding a barycenter in the space of probability distributions is solving an optimization problem that results in alternating descent to find vectors v,w. This makes finding the 'mean' of distributions a question of minimizing a mix KL Divergences, which results in an iterative solution using a Sinkhorn method.

May implement for 3d voxels in GL and CUDA.

About

Convolutional Wasserstein Distances, in Python, based on a lot of peoples work.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages