BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Relations and Predictions in Brains and Machines - Dr. Kim Stachen
 feld\, Deep Mind
DTSTART:20230302T150000Z
DTEND:20230302T160000Z
UID:TALK195394@talks.cam.ac.uk
CONTACT:Sofia Orellana
DESCRIPTION:Humans and animals learn and plan with flexibility and efficie
 ncy well beyond that of modern Machine Learning methods. This is hypothesi
 zed to owe in part to the ability of animals to build structured represent
 ations of their environments\, and modulate these representations to rapid
 ly adapt to new settings. In the first part of this talk\, I will discuss 
 theoretical work describing how learned representations in hippocampus ena
 ble rapid adaptation to new goals by learning predictive representations\,
  while entorhinal cortex compresses these predictive representations with 
 spectral methods that support smooth generalization among related states. 
 I will also cover recent work extending this account\, in which we show ho
 w the predictive model can be adapted to the probabilistic setting to desc
 ribe a broader array of generalization results in humans and animals\, and
  how entorhinal representations can be modulated to support sample generat
 ion optimized for different behavioral states. In the second part of the t
 alk\, I will overview some of the ways in which we have combined many of t
 he same mathematical concepts with state-of-the-art deep learning methods 
 to improve efficiency and performance in machine learning applications lik
 e physical simulation\, relational reasoning\, and design.
LOCATION:Online
END:VEVENT
END:VCALENDAR