Wasserstein Dictionary learning and Wasserstein NMF

Relevant paper: Antoine Rolet, Marco Cuturi and Gabriel Peyré. Fast Dictionary Learning with a Smoothed Wasserstein Loss.. To appear in Proceedings of the 19th International Conference on Artificial Intelligence and Statistics (AISTATS) 2016, Cadiz, Spain. JMLR: W&CP volume 41. Supplementary materials.

Code available on the Github repository: http://arolet.github.io/wasserstein-dictionary-learning


Japanese Optical Character Recognition / Dictionary app, available on iOS and Android:

Available on the App Store Get it on Google Play

Transportation distance

Here is a matlab wrapper for the c++ code from Nicolas Bonneel available on this page, which is a light adaptation of LEMON to solve the network simplex for bipartite graphs. I recommend using the source code, but you can also use one of the system specific compiled matlab function.

Update 2015-05:

  • Corrected a bug that appears for highly degenerated problem that made the optimality check fail (leading to infinite cycling)
  • Added an argument that allows to limit the maximum number of simplex pivot (defaults to infinity)

Source code

Mac :

Linux :

Windows :

  • Coming soon