BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Computational Neuroscience Journal Club - Dylan Festa (University 
 of Cambridge)
DTSTART:20140128T160000Z
DTEND:20140128T170000Z
UID:TALK50571@talks.cam.ac.uk
CONTACT:Guillaume Hennequin
DESCRIPTION:Dylan Festa will present:\n\nPredictive Coding of Dynamical Va
 riables in Balanced Spiking Networks\nby Martin Boerlin\, Christian K. Mac
 hens and Sophie Denève\nPLoS Comp. Biol. (2013)\n\nhttp://www.ploscompbio
 l.org/article/info%3Adoi%2F10.1371%2Fjournal.pcbi.1003258\n\nAbstract\n\nT
 wo observations about the cortex have puzzled neuroscientists for a long t
 ime. First\, neural responses are highly variable. Second\, the level of e
 xcitation and inhibition received by each neuron is tightly balanced at al
 l times. Here\, we demonstrate that both properties are necessary conseque
 nces of neural networks that represent information efficiently in their sp
 ikes. We illustrate this insight with spiking networks that represent dyna
 mical variables. Our approach is based on two assumptions: We assume that 
 information about dynamical variables can be read out linearly from neural
  spike trains\, and we assume that neurons only fire a spike if that impro
 ves the representation of the dynamical variables. Based on these assumpti
 ons\, we derive a network of leaky integrate-and-fire neurons that is able
  to implement arbitrary linear dynamical systems. We show that the membran
 e voltage of the neurons is equivalent to a prediction error about a commo
 n population-level signal. Among other things\, our approach allows us to 
 construct an integrator network of spiking neurons that is robust against 
 many perturbations. Most importantly\, neural variability in our networks 
 cannot be equated to noise. Despite exhibiting the same single unit proper
 ties as widely used population code models (e.g. tuning curves\, Poisson d
 istributed spike trains)\, balanced networks are orders of magnitudes more
  reliable. Our approach suggests that spikes do matter when considering ho
 w the brain computes\, and that the reliability of cortical representation
 s could have been strongly underestimated.\n\n
LOCATION:Cambridge University Engineering Department\, CBL Rm #438 (http:/
 /learning.eng.cam.ac.uk/Public/Directions)
END:VEVENT
END:VCALENDAR