Open jiwoncpark opened 4 years ago
Thanks @jiwoncpark ! Through a mixture of business and absent-mindedness we seem to have skipped a meeting, but the calendar tells me that we are due to meet next week, Tuesday March 10. How about we take a look at this paper you suggest next time? What do you think, everyone?
Slides here
Hi All, will you be setting up a ZOOM?
Rezende and Mohamed 2015 is a classic paper on using normalizing flows, rather than e.g. mean-field approximations, for variational inference. Normalizing flows are appropriate for problems with degeneracies and outliers, as they represent a more flexible family of posterior distributions as compared with factorized Gaussians. Flow-based models were also heavily endorsed by Francois at the Likelihood-free Inference Workshop! They deserve mention in our DL lit review, I think.
Possibly interesting keywords: RealNVP, NICE, Glow (models with normalizing flows) MADE, PixelRNN, WaveNet, MAF, IAF (models with autoregressive flows)