BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Computational Neuroscience Journal Club - Dr. David Barrett (Unive
 rsity of Cambridge)
DTSTART:20141104T160000Z
DTEND:20141104T170000Z
UID:TALK55952@talks.cam.ac.uk
CONTACT:Guillaume Hennequin
DESCRIPTION:David Barrett will cover: Stochastic variational learning in r
 ecurrent spiking networks\, Rezende D and Gerstner W\, Frontiers in Comput
 ational Neuroscience 2014 (http://journal.frontiersin.org/Journal/10.3389/
 fncom.2014.00038/abstract).\n\nABSTRACT: The ability to learn and perform 
 statistical inference with biologically plausible recurrent networks of sp
 iking neurons is an important step toward understanding perception and rea
 soning. Here we derive and investigate a new learning rule for recurrent s
 piking networks with hidden neurons\, combining principles from variationa
 l learning and reinforcement learning. Our network defines a generative mo
 del over spike train histories and the derived learning rule has the form 
 of a local Spike Timing Dependent Plasticity rule modulated by global fact
 ors (neuromodulators) conveying information about "novelty" on a statistic
 ally rigorous ground. Simulations show that our model is able to learn bot
 h stationary and non-stationary patterns of spike trains. We also propose 
 one experiment that could potentially be performed with animals in order t
 o test the dynamics of the predicted novelty signal.
LOCATION:Cambridge University Engineering Department\, CBL\, BE-438 (http:
 //learning.eng.cam.ac.uk/Public/Directions)
END:VEVENT
END:VCALENDAR