TSNE different way

tsne is a important method to plot high dimensional figure, and many people
bulid different way to realise it. i will introduce some methods recently use.

1. sklearn

sklearn is a very popular package to slove machine learning problem

The disadvantage is that only sinlge kernal can be used.

from sklearn.manifold import TSNE
model = TSNE(n_components=2,init='pca')
node_pos = model.fit_transform(X)

2. openTSNE
   opentsne can use multi worker to do the job,
   https://opentsne.readthedocs.io/en/latest/examples/01/01_simple_usage.html

from openTSNE import TSNE
model = TSNE(n_components=2,init='pca',n_jobs=8,)
node_pos = model.fit(X)

3. tsne-cuda
   tsne-cuda can use cude to accelerate tsne, but the installation is complex.
   https://github.com/CannyLab/tsne-cuda