metric-learn

metric-learn copied to clipboard

Dimensionality reduction for algorithms learning Mahalanobis matrix M

3

For all Mahalanobis metric learners, we should be able to reduce the dimension. For those that optimize the transformation matrix `L`, this can be done explicitely by setting the matrix...

wdevazelhes

doctests not working

1

In PR #152 I'll introduce some doctests. However, they seem to not be executed at least when I build the docs (if I change the results it throws no error...

wdevazelhes

Towards possible M2TML implem

2

Hi `metric-learn` team ! We have [discussed recently with `tslearn`](https://github.com/rtavenar/tslearn/issues/8) on the possibility of implementing [M2TML](https://tel.archives-ouvertes.fr/tel-01678889v1/document) metric learning, which generalizes the "large margin" concepts in LMNN (Weinberger and Saul) in...

smarie

Add a get_statistics() method

2

Hi, Is there any roadmap for adding a get_statistics() function, like the one that Shogun offers, which helps us to get the statistics of a certain model? It could be...

anirudt

Implementing Incremental Fit for LMNN

3

Currently, each time the `fit` method is called, it starts from a new identity matrix. But in many scenarios, it can be useful to incrementally fit (i.e. start from the...

sidak

Warning on RCA with chunk labeling not starting on zero or with gaps

2

#### Description RCA chunks are expected to start at zero and increase one by one, this raises a warning in case it doesn't start at zero or has any gap....

grudloff

Adds a semi-supervised (specifically a combination of supervised and weakly-supervised data) version of weak algorithms

4

Closes #233 For now I only wrote what I believe to be expected for #233 for the RCA algorithm. It is a simple modification of the supervised version of the...

RobinVogel

Combination of supervised and weakly-supervised data

2

We could easily allow users to fit weakly-supervised algorithms on a combination of label supervision (from which we generate constraints as in supervised versions) and additional weak supervision specified by...

bellet

TupleTransformer

2

As discussed with @bellet, it would be useful to have a sort of `TupleTransformer` object, that would take as `__init__` a regular scikit-learn `Transformer` (so it would be a `MetaEstimator`),...

wdevazelhes

I just realized from here https://scikit-learn.org/stable/modules/cross_validation.html#computing-cross-validated-metrics, that "When the cv argument is an integer, cross_val_score uses the KFold or StratifiedKFold strategies by default, the latter being used if the estimator...

wdevazelhes