BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Computational Neuroscience Journal Club - Greg Sotiropoulos (CBL)
DTSTART:20180123T160000Z
DTEND:20180123T170000Z
UID:TALK99112@talks.cam.ac.uk
CONTACT:Rodrigo Echeveste
DESCRIPTION:Greg Sotiropoulos will cover:\n\n• Toward a unified theory o
 f efficient\, predictive\, and sparse coding\n\n\n• Matthew Chalk\, Oliv
 ier Marre\, and Gašper Tkačik\n\n• PNAS (2017)\n\n• http://www.pnas.
 org/content/115/1/186.abstract\n\n\nAbstract: A central goal in theoretica
 l neuroscience is to predict the response properties of sensory neurons fr
 om first principles. To this end\, “efficient coding” posits that sens
 ory neurons encode maximal information about their inputs given internal c
 onstraints. There exist\, however\, many variants of efficient coding (e.g
 .\, redundancy reduction\, different formulations of predictive coding\, r
 obust coding\, sparse coding\, etc.)\, differing in their regimes of appli
 cability\, in the relevance of signals to be encoded\, and in the choice o
 f constraints. It is unclear how these types of efficient coding relate or
  what is expected when different coding objectives are combined. Here we p
 resent a unified framework that encompasses previously proposed efficient 
 coding models and extends to unique regimes. We show that optimizing neura
 l responses to encode predictive information can lead them to either corre
 late or decorrelate their inputs\, depending on the stimulus statistics\; 
 in contrast\, at low noise\, efficiently encoding the past always predicts
  decorrelation. Later\, we investigate coding of naturalistic movies and s
 how that qualitatively different types of visual motion tuning and levels 
 of response sparsity are predicted\, depending on whether the objective is
  to recover the past or predict the future. Our approach promises a way to
  explain the observed diversity of sensory neural responses\, as due to mu
 ltiple functional goals and constraints fulfilled by different cell types 
 and/or circuits. 
LOCATION:Cambridge University Engineering Department\, CBL\, BE4-38 (http:
 //learning.eng.cam.ac.uk/Public/Directions)
END:VEVENT
END:VCALENDAR