Thread by @leland_mcinnes

The latest version of umap-learn is now out. Version 0.5 includes some major new features, including ParametricUMAP, DensMAP, AlignedUMAP, model composition, and model updating. Thank you to everyone who contributed! 1/14

ParametricUMAP uses a neural network to learn a UMAP embedding. This allows for a number of significant advantages. 2/14

ParametricUMAP provides extremely fast new data embedding (comparable to PCA if you use a GPU), UMAP based autoencoders, and powerful semi-supervised learning, particularly in low label regimes. 3/14

Special thanks got to @tim_sainburg who contributed the ParametricUMAP implementation. See the documentation (umap-learn.readthedocs.io/en/latest/para…), or the paper (arxiv.org/abs/2009.12981), for more details. 4/14

DensMAP is a modification of UMAP that provides better preservation of relative local density. See the documentation (umap-learn.readthedocs.io/en/latest/dens…) or the paper (biorxiv.org/content/10.110…) for further details. 5/14

Special thanks to Hyunghoon Cho and team for their contribution of DensMAP based on their paper. 6/14

AlignedUMAP allows sequences of different UMAP embeddings to be aligned with each other according to relations among the datasets. This can be particularly useful for situations such as time evolving data. 7/ 14

See the documentation (umap-learn.readthedocs.io/en/latest/alig… and umap-learn.readthedocs.io/en/latest/alig…) for more details and examples of AlignedUMAP. 8/14

UMAP also now supports some degree of model composition, allowing users to combine different UMAP models by

model1 * model2
model1 + model2
model1 - model2

See the documentation for more details (umap-learn.readthedocs.io/en/latest/comp…). 9/14

UMAP also now allows use of an “update” method to generate a new model updated with new additional data. 10/14

The approximate nearest neighbour search from UMAP is now fully moved to an ANN library PyNNDescent (github.com/lmcinnes/pynnd…). In turn PyNNDescent has seen significant development and is faster, multithreaded, and supports new metrics such as Wasserstein distance. 11/14

Finally a large number of bug-fixes, plotting improvements, and performance improvements were contributed as well. 12/14

Thank you to *everyone* who contributed, including those who helped in improving the documentation. 13/14

To upgrade to the new umap-learn you can use:

pip install umap-learn —upgrade

or

conda update umap-learn

to freshly install the new umap-learn:

pip install umap-learn

or

conda install umap-learn

or install from the source on GitHub (github.com/lmcinnes/umap). 14/14