Let’s Try t-SNE

Re:Let’s Try t-SNE by b_tterc_p on Sunday December 23, 2018 @12:07AM (#18744553)

To anyone using t sne, consider giving UMAP a try.

Equally accessible as tsne in Python.

[ Parent | on HN ]

Re:Let’s Try t-SNE by aflam on Sunday December 23, 2018 @08:19AM (#18745909)

Thanks for the article.I came across it as part of scikit-tda, but did not investigate. Looks promising! The TDA background is appealing, as well as runtime complexity.

[ Parent | on HN ]

Re:Let’s Try t-SNE by amelius on Sunday December 23, 2018 @09:49AM (#18746327)

In this tutorial the author reduces the size of the images from 28x28 to 10x10 to make the computation more tractable.

Would that be necessary with UMAP too?

[ Parent | on HN ]

Re:Let’s Try t-SNE by b_tterc_p on Tuesday December 25, 2018 @12:49AM (#18756151)

T SNE and UMAP are techniques for dimensionality reduction. Both of them scale pretty well to high dimensionality problems. I think the author downsized the data because he wanted to provide an interactive model. If I were doing this I would probably just leave it as is for either method. The article I linked actually does benchmark on the same dataset though. For the full 70,000 images at 28x28, it claims 87 seconds for UMAP and 1450 for TSNE.

[ Parent | on HN ]

Re:Let’s Try t-SNE by SubiculumCode on Sunday December 23, 2018 @02:54AM (#18744981)

Is there an R implementation?

[ Parent | on HN ]

Re:Let’s Try t-SNE by amrrs on Sunday December 23, 2018 @04:13AM (#18745202)

Yes https://www.kaggle.com/nulldata/tsne-alternate-umap-3d-viz-o... this has links to R UMAP implementation

[ Parent | on HN ]