[Adapt] seminar topic: t-SNE

[新青年]

Hi, Adapters:
  I will introduce a manifold learning method t-SNE, which can be regarded as dimensionality  reduction ,I think.
 
  There exist many methods about  dimensionality  reduction, some linear like Principal Component Analysis(PCA), Linear Discriminant Analysis(LDA)…, some nonlinear like Isomap, LoLaplaciancally Linear Embedding(LLE)…. and I will introduce one nonlinear method t-SNE, by which The visualizations produced are significantly better than those produced by the other techniques on almost all of the datasets.
 
   The idea of this algorithm is clear, But there still exist many details deserved to be understood. I hope you can have some review of probability theory before, such as Gaussian distribution, conditional probability, and t-distribution. And KL divergence, which is also called relative entropy, is also needed.
 
 Here  are some useful references ordered by importance:
 
Maaten L, Hinton G. Visualizing data using t-SNE[J]. Journal of Machine Learning Research, 2008, 9(Nov): 2579-2605.

http://bindog.github.io/blog/2016/06/04/from-sne-to-tsne-to-largevis

http://blog.pluskid.org/?p=533

Best Wishes

Haijun
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cs.sjtu.edu.cn/pipermail/adapt/attachments/20170314/b6a14cd5/attachment.html>