The Case for Bayesian Deep Learning | Andrew's Blog

6 bookmarks. First posted by jimcmcdonald 16 days ago.

Bayesian inference is especially compelling for deep neural networks. The key distinguishing property of a Bayesian approach is marginalization instead of optimization, not the prior, or Bayes rule. Neural networks are typically underspecified by the data, and can represent many different but high performing models corresponding to different settings of parameters, which is exactly when marginalization will make the biggest difference for accuracy and calibration. Moreover, deep ensembles can be seen as approximate Bayesian marginalization.

bayesian
ai
ml
15 days ago by Hwinkler

I have written a note on the case for Bayesian deep learning:

from twitter_favs
16 days ago by knowlsie