Stochastic Neighbor Embedding with Gaussian and Student-t Distributions: Tutorial and Survey

Stochastic Neighbor Embedding (SNE) is a manifold learning and dimensionality\nreduction method with a probabilistic approach. In SNE, every point is consider\nto be the neighbor of all other points with some probability and this\nprobability is tried to be preserved in the embedding space. SNE considers\nGaussian distribution for the probability in both the input and embedding\nspaces. However, t-SNE uses the Student-t and Gaussian distributions in these\nspaces, respectively. In this tutorial and survey paper, we explain SNE,\nsymmetric SNE, t-SNE (or Cauchy-SNE), and t-SNE with general degrees of\nfreedom. We also cover the out-of-sample extension and acceleration for these\nmethods.\n