BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Local and global synaptic credit assignment - Dr Rui Ponte Costa (
 University of Bristol)
DTSTART:20190118T110000Z
DTEND:20190118T120000Z
UID:TALK117607@talks.cam.ac.uk
CONTACT:Rodrigo Echeveste
DESCRIPTION:For animals to learn to interact with their environments milli
 ons of synapses across several brain areas need to be modified appropriate
 ly. Understanding how the brain assigns credit and modifies a particular s
 ynapse is a long standing problem in neuroscience.\nIn this talk I will st
 art by describing our work on a theoretical framework of how synapses may 
 (locally) assign credit to their response parameters to achieve a desired 
 target. Such target responses may\, in turn\, be computed by gradient desc
 ent algorithms that compare the network output with the desired outcome. I
 n machine learning\, the backpropagation algorithm is the prime example of
  one such solution which solves the global synaptic credit assignment prob
 lem. I will finish by discussing recent work proposing that cortical micro
 circuits with dendritic properties may indeed approximate backprop. Overal
 l\, our work provides a fresh perspective on how the brain may solve the s
 ynaptic credit assignment problem.
LOCATION:Cambridge University Engineering Department\, CBL\, BE4-38 (http:
 //learning.eng.cam.ac.uk/Public/Directions)
END:VEVENT
END:VCALENDAR