fugw.mappings#
Most likely, this is the module you want to use.
It comprises 2 main classes: fugw.mappings.FUGW
and fugw.mappings.FUGWSparse,
each of which comes with .fit(), transform() and .inverse_transform()
methods to easily fit your FUGW models and apply them to new data.
Warning
Mappings overwrite __getstate__() and __setstate__() methods
in order to store the model hyper-parameters and weights separately in the same
pickle files. This allows for loading hyper-parameters only, which results in
much faster loading times.
However, this means you should be careful when
saving or loading mappings with pickle. If you want to save/load a mapping, use
fugw.utils.save() and fugw.utils.load() methods instead.
|
Class computing dense transport plans. |
|
Class computing sparse transport plans |