BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Frugal computation - Xaq Pitkow (CMU)
DTSTART:20260122T143000Z
DTEND:20260122T160000Z
UID:TALK243283@talks.cam.ac.uk
CONTACT:Daniel Kornai
DESCRIPTION:The brain implements algorithms that choose smart actions to a
 chieve its goals\, using a tiny fraction of the power used by today’s co
 mputers on comparable tasks. Modeling these algorithms requires us to acco
 unt not only for the task demands\, but also for the costs of thinking. We
  call this problem setting “frugal computation”. I will describe two s
 tudies that incorporate computational costs into control problems. Specifi
 cally\, we generalize past work on efficient coding and predictive coding\
 , by accounting for either the representational costs of integrating infor
 mation or the computational costs of performing that integration. We find 
 that the predictability of the system determines phase transitions between
  strategies\, showing when it is worth spending computational resources to
  achieve better performance.
LOCATION:CBL Seminar Room\, Engineering Department\, 4th floor Baker build
 ing
END:VEVENT
END:VCALENDAR