BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Cambridge ELLIS Seminar Series- Dr David Barber- Training Energy B
 ased Models - Speaker to be confirmed
DTSTART:20231023T140000Z
DTEND:20231023T150000Z
UID:TALK206371@talks.cam.ac.uk
CONTACT:Kimberly Cole
DESCRIPTION:The Cambridge ELLIS Unit Seminar Series holds talks by leading
  researchers in the area of\nmachine learning and AI. Our next speaker for
  2023 will be Dr. David Barber. Details of his\ntalk can be found below.\n
 \nTitle: “Training Energy Based Models"\n\nAbstract: Energy Based models
  are undirected models of the formp(x) \\propto exp(-f(x)). These are very
  general models and provide an alternative to the currently popular direct
 ed models used in Diffusion Models and LLMs. However\, training and sampli
 ng from these models is typically problematic. We consider an approach bas
 ed on Denoising Score Matching and consider a Gibbs sampling to recover sa
 mples from the learned model. Remarkably\, we show that there is an exact 
 expression for the mean and covariance of the sampler. We will also\nhighl
 ight close connections to Diffusion Models.\n\nhttps://cam-ac-uk.zoom.us/j
 /83278351798?pwd=bUh6d09NVkkwM20xeFVXZUVjakNTQT09
LOCATION:Zoom
END:VEVENT
END:VCALENDAR