block by Fil eaee5850dfbb30d7e326722c7bca7fac

t-SNE and Urquhart w/distances study

Full Screen

This is a t-SNE representation of an array of (random) circles : red, green, blue, opacity and radius are 5 independent dimensions.

The t-SNE is computed by a javascript worker, using science.ai’s implementation.

A force is created that tries to:

The result is displayed with an Urquhart graph.

On the left: the Urquhart graph eliminates links that have the longest projected distances;
On the right: using the original distances. (This algorithm doesn’t connect dots just because they’re projected together.)

I find it interesting that sites that are far apart (on the edge of the graph) are closer in the original space than it seems on the projected space. Also, closely related colours in the center are more connected.

Forked from t-SNE, a force, and voronoi.

index.html

worker.js