Rémi Flamary

[WIP] New alpha API for OT solver (with pre-computed ground cost matrix)

4

## Types of changes I implement the new POT API for general OT solvers. It comes with the new function `ot.solve` that can be used to solve exact, regularized and...

[WIP] Gaussian initialization for sinkhorn

3

## Types of changes This is a first shot for intializing empirical sinkhorn but the computational gain is not very clear on tests I did locally. The following code ```python...

[WIP] Make `benchopt.cli`an executable module

1

This PR aim at alowing calling benchopt from any python interpreter with something like ``` python -m benchopt.cli ``` Basically we just need to add a file in the cli...

It would be very interesting to add an option to also plot a table explorer for the results on the web page.

## Types of changes ## Motivation and context / Related issue ## How has this been tested (if it applies) ## PR checklist - [ ] I have read the...

## Describe the bug We needed to force an older version of jax to prevent a bug when importing pymanopt because of a very recent deprecation of the configuration setup...

[WIP] Add multi-domain Monge alignment and JCPOT Target shift method

1

Adds a multi domain Monge alignement Adapter to SKADA that does the following 1. compte a barycenter of all sources (Gaussian assumtion) 2. Map all data to this barycenter with...

While this feature makes no sens for fitting, it can be very practical for multi-source/target for instance it would allow to sometging like ```python pipe = make_da_pipeline( PerDomain(StandardScaler()), SelectSource(SVC()), )...

Merge all OT mapping methods in one class

2

Should we merge OT mapping methods? it seems like the use of entropic regularization or group lasso could be added easily as a parameter to the same class. We should...

https://projet.liris.cnrs.fr/imagine/pub/proceedings/CVPR2012/data/papers/261_O2D-03.pdf