Over the past couple of weeks I have spent a lot of time reading up on dimensionality reduction papers, ranging from PCA, random projections, LLEs, Isomaps, Conformal eigenmaps, etc. LLEs really struck me as interesting and I came across an introductory paper that really gave a nice, easy to understand overview. Page 7 of the intro paper shows how LLEs can map images with corner faces to the corners of the two dimensional embedding. I understand how this works, but I am really struggling to mimic their visualization. If anyone can point me in the right direction or even has some sample code I would be really appreciative. I really want to start playing around with dimensionality reduction visualizations, but I'm honestly spinning my wheels right now.
[link] [6 comments]