P2333/Max-Mahalanobis-Training

the configuration of t-SNE visualization

sndnyang opened this issue · 2 comments

sndnyang commented

In your paper, you have very dense clusters in your t-SNE visualization plot.

image

I can generate very dense logits in 10-dimension to the true center.
the features/outputs after the LDA like [100+, 100+, 100+, 0.1-, 100+, 100+, ...]. It's very close to the center in 10d.
But when I apply t-SNE (with sklearn library), it generates something like sticks/bands.

image

So any suggestion to generate the t-SNE visualization better as yours?

sndnyang commented

OK, I made mistakes. Solve it. I should use the features, not logits~~~

sndnyang commented

OK, I made mistakes. Solve it. I should use the features, not logits~~~