684 karma · joined March 1, 2016
So, each video is mapped to fixed size vector of floats? A user's history is now a matrix of size [number of videos, embedding size]? What are the other parameters in this sentence "Importantly, the embeddings are learned jointly with all other model parameters through normal gradient descent back propagation updates."? And how do you concatenate all these into a "wide layer" when users would have histories of different length?
They do have the technical talents though. I think one of the word2vec authors was there, but he was giving a talk about using the same technique on emoji icons :-( I would think there are more interesting problems to tackle with the available resources at FB.
It's really Theranoax.
http://www.nytimes.com/2016/04/19/business/theranos-sec-just...
In early October 2013, Guillory met with 28-year-old Vogt, a self-proclaimed MIT drop- out who had spent a month to earn a degree in installing Microsoft Windows.