It's time to stop making t-SNE & UMAP plots. In a new preprint w/ Tara Chari we show that while they display some correlation with the underlying high-dimension data, they don't preserve local or global structure & are misleading. They're also arbitrary.🧵


It turns out you can't squeeze a high-dimensional structure in 2D without arbitrary distortions. Shock and horror.

Important, and constructive work. But I wish it was not needed in the first place!

To the poor souls still doing visual network analysis based on 2D spring-block embedding (a.k.a. "ridiculograms"): please read this paper very carefully, and despair!

Rir pra não chorar ... ( & valeu pela referência)
Sign in to participate in the conversation
Mastodon @

The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!