Poincare distance to origin

[ruoshiliu]

Thanks a lot for sharing the repo! In your paper, you presented a nice visualization showing the distribution of hyperbolic distances of embeddings (MNIST vs. Omniglot). I'm wondering from which stage did you obtain the embeddings to calculate those distances?

My guess is the output of the line below. Can you please confirm this is indeed the case?

hyperbolic-image-embeddings/examples/mnist.py

Line 34 in 05b104e

x = self.tp(x)

Thanks!

[KhrulkovV]

Hello! Yes, this is exactly right. self.tp(x) applied to features x in the Euclidean space outputs hyperbolic features (a tensor of the same shape with rows lying inside of the Poincare ball).

[ruoshiliu]

No problem, thanks!

[ruoshiliu]

Hi Valentin, please forgive my ignorance here. Can you please explain more about how you generated the poincare ball visualization with MNIST in your paper? Specifically, is that D^2 projected to R^2 with pmath.logmap0? What does the radius represent?