metric-learn
metric-learn copied to clipboard
scikit-learn-contrib
Metric learning algorithms in Python
Methods that we haven't implemented yet, but would like to. In no particular order: - [ ] [Probablistic Global Distance Metric Learning](http://www.cs.cmu.edu/~liuy/frame_survey_v2.pdf) - Probabilistic extension of MMC (pMMC?) - [...
We need to ensure that as much as possible no warning is raised for the user (for instance by running all the tests and checking what warning are raised)
In the auto init, there is a misleading behaviour that needs to be addressed: if I specify `init='lda'` , with a number of components of n_features, (and I have n_classes...
Tests take a significant time to run, especially tests that need to run on datasets that have a reasonable size. Possible options to make the tests go faster are: -...
Just a thought, but it might be useful to have a function to update/set the preprocessor outside of the init. Right now it is just initialized when calling fit (so...
#### Description Hi, is there going to be some metric learning algorithm on the semi-supervised direction, utilising both labels/pairwise constraints and unlabelled data to derive the distance metric. #### Some...
I just realized for SDML, we raise the error that would be returned in the graphical lasso but if it raises no error but returns a non finite matrix, then...
Related to #168 scikit-learn's `CalibratedClassifierCV` should work with pairs classifier: We should ensure that it return coherent results with our estimators (could be done in #168) This could be finalized...
#### Description ITML raises an error on this example (1D iris). We should investigate why, and see if it is the surface of some problem or not Note that if...
#### Description LSML returns an error in some particular cases. We should decide what to do in those cases, either let it as such, or return a particular error. This...
Metadata
Owner
Metadata
Metric learning algorithms in Python