BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Bayesian Neural Networks - James Allingham\, Javier Antorán\, Vin
 cent Fortuin (University of Cambridge)
DTSTART:20230222T110000Z
DTEND:20230222T123000Z
UID:TALK197725@talks.cam.ac.uk
CONTACT:James Allingham
DESCRIPTION:Bayesian Neural Networks (BNNs) take a probabilistic approach 
 to learning in neural networks by placing distributions over the weights a
 nd performing (approximate) Bayesian inference. In this talk\, we will int
 roduce the basics of BNNs\, the challenges in training them\, and some of 
 their properties. We will then discuss the Laplace approximation as a high
 ly performant approximate inference scheme for BNNs with connections to li
 near models and neural tangent kernels. Finally\, we will provide a tour o
 f prior choices in BNNs\, looking at both weight and function space approa
 ches.\n\nRecommended reading:\n\nMacKay\, D. J. (1992). Bayesian interpola
 tion. Neural computation\, 4(3)\, 415-447. https://authors.library.caltech
 .edu/13792/1/MACnc92a.pdf \n\nBlundell\, C.\, Cornebise\, J.\, Kavukcuoglu
 \, K.\, & Wierstra\, D. (2015\, June). Weight uncertainty in neural networ
 k. In International conference on machine learning (pp. 1613-1622). PMLR. 
 http://proceedings.mlr.press/v37/blundell15.html\n\nImmer\, A.\, Korzepa\,
  M.\, & Bauer\, M. (2021\, March). Improving predictions of Bayesian neura
 l nets via local linearization. In International Conference on Artificial 
 Intelligence and Statistics (pp. 703-711). PMLR. http://proceedings.mlr.pr
 ess/v130/immer21a.html\n\nFortuin\, V. (2022). Priors in bayesian deep lea
 rning: A review. International Statistical Review\, 90(3)\, 563-591. https
 ://onlinelibrary.wiley.com/doi/pdfdirect/10.1111/insr.12502
LOCATION:Cambridge University Engineering Department\, CBL Seminar room BE
 4-38.
END:VEVENT
END:VCALENDAR