News for hackers. Stuff that matters.
Re:Machine learning works spectacularly well, but mathematic... by jeremysalwen on Saturday December 05, 2015 @02:10AM (#10681177)
This reminded me of this paper on using deep autoencoders as a generative model: http://papers.nips.cc/paper/5023-generalized-denoising-auto-... I don't think it's exactly a right analogy, but there's something there about the local flow of MCMC locally describing the manifold structure, and thus generating it globally.

There are no comments on this comment yet.