NEMLD 2013

The New England Machine Learning Day @ the Microsoft NERD Center was today. Below is the breakdown of the talks. Best technical talk? Ryan Adams. Funniest talk? Antonio Torralba. Sham Kakade (MSR) Learning latent structure in documents, social networks, and…

Motion amplification for the masses

One of my first classes at MIT was the "Advances in Computer Vision" course. It was both well-taught and incredibly informative, likely because Dr. Freeman/Dr. Torralba/Michael Rubinstein were so knowledgeable about the content. There was an article recently on one…

PCA via SVD

I was recently asked how singular value decompostion (SVD) could be used to perform principal component analysis (PCA). SVD is a general matrix decomposition method that can be used on any m × n matrix. (Compare this to eigenvalue decomposition,…