Animation generation for scikit-learn's t-sne algorithm
Strong advice - use tensorboard, it's not so complicated.
But if you want dirty quick visualisation - here we go.
Requirements:
ffmpeg or mencoder to save gifs
Easy to install:
$ pip install tsne_animate
Easy to use:
from sklearn import datasets, manifold
from tsne_animate import tsneAnimate
digits = datasets.load_digits()
tsne = tsneAnimate(manifold.TSNE(learning_rate=1000))
tsne.animate(digits.data,digits.target)
from sklearn import datasets, manifold
from tsne_animate import tsneAnimate
iris = datasets.load_iris()
tsne = tsneAnimate(manifold.TSNE(learning_rate=50))
tsne.animate(iris.data,iris.target,0,'iris.gif')