We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Statistics > Machine Learning

Title: A statistical theory of cold posteriors in deep neural networks

Abstract: To get Bayesian neural networks to perform comparably to standard neural networks it is usually necessary to artificially reduce uncertainty using a "tempered" or "cold" posterior. This is extremely concerning: if the prior is accurate, Bayes inference/decision theory is optimal, and any artificial changes to the posterior should harm performance. While this suggests that the prior may be at fault, here we argue that in fact, BNNs for image classification use the wrong likelihood. In particular, standard image benchmark datasets such as CIFAR-10 are carefully curated. We develop a generative model describing curation which gives a principled Bayesian account of cold posteriors, because the likelihood under this new generative model closely matches the tempered likelihoods used in past work.
Comments: Published at ICLR 2021 (this https URL)
Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG)
Cite as: arXiv:2008.05912 [stat.ML]
  (or arXiv:2008.05912v2 [stat.ML] for this version)

Submission history

From: Laurence Aitchison [view email]
[v1] Thu, 13 Aug 2020 13:46:58 GMT (16kb)
[v2] Tue, 27 Apr 2021 14:33:30 GMT (2733kb,D)

Link back to: arXiv, form interface, contact.