BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Circuits and Interpretability - Lauro Langosco\, Elre Oldewage and
  Juyeon Heo(University of Cambridge)
DTSTART:20220216T110000Z
DTEND:20220216T123000Z
UID:TALK169040@talks.cam.ac.uk
CONTACT:Elre Oldewage
DESCRIPTION:In this talk we will look at methods that aim to make the inte
 rnal computations of neural networks visible (‘interpretable’) to huma
 ns. This is useful for a) making deep learning models robust / fair / safe
  and b) in order to come to an empirical\, scientific understanding of why
  deep learning works. We will cover various methods from the literature\, 
 and focus in particular on the study of circuits\, i.e. modular subnetwork
 s that serve a particular function. \n\n*Recommended reading:*\n\nThe Buil
 ding Blocks of Interpretability (https://distill.pub/2018/building-blocks/
 )\n\n*Optional reading:*\n\n# Adversarial Examples Are Not Bugs\, They Are
  Features (https://arxiv.org/abs/1905.02175)\n# Are Neural Nets Modular? I
 nspecting Functional Modularity Through Differentiable Weight Masks (https
 ://arxiv.org/abs/2010.02066)\n\nOur reading groups are live-streamed via Z
 oom and recorded for our Youtube channel. The Zoom details are distributed
  via our weekly mailing list.
LOCATION: Cambridge University Engineering Department \,LR3A
END:VEVENT
END:VCALENDAR