BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Computational Neuroscience Journal Club - Thomas Burger (Control G
 roup)
DTSTART:20190528T150000Z
DTEND:20190528T160000Z
UID:TALK125677@talks.cam.ac.uk
CONTACT:Rodrigo Echeveste
DESCRIPTION:Thomas Burger will present:\n\n• Learning Nonlinear Dynamics
  in Efficient\, Balanced Spiking Networks Using Local Plasticity Rules\n\n
 • Alireza Alemi\, Christian K. Machens\, Sophie Deneve\, Jean-Jacques Sl
 otine\n\n• AAAI Conference on Artificial Intelligence (2018)\n\n• http
 s://aaai.org/ocs/index.php/AAAI/AAAI18/paper/view/17438\n\n\nAbstract: The
  brain uses spikes in neural circuits to perform many dynamical computatio
 ns. The computations are performed with properties such as spiking efficie
 ncy\, i.e. minimal number of spikes\, and robustness to noise. A major obs
 tacle for learning computations in artificial spiking neural networks with
  such desired biological properties is due to lack of our understanding of
  how biological spiking neural networks learn computations. Here\, we cons
 ider the credit assignment problem\, i.e. determining the local contributi
 on of each synapse to the network's global output error\, for learning non
 linear dynamical computations in a spiking network with the desired proper
 ties of biological networks. We approach this problem by fusing the theory
  of efficient\, balanced neural networks (EBN) with nonlinear adaptive con
 trol theory to propose a local learning rule. Locality of learning rules a
 re ensured by feeding back into the network its own error\, resulting in a
  learning rule depending solely on presynaptic inputs and error feedbacks.
  The spiking efficiency and robustness of the network are guaranteed by ma
 intaining a tight excitatory/inhibitory balance\, ensuring that each spike
  represents a local projection of the global output error and minimizes a 
 loss function. The resulting networks can learn to implement complex dynam
 ics with very small numbers of neurons and spikes\, exhibit the same spike
  train variability as observed experimentally\, and are extremely robust t
 o noise and neuronal loss.\n
LOCATION:Cambridge University Engineering Department\, CBL\, BE4-38 (http:
 //learning.eng.cam.ac.uk/Public/Directions)
END:VEVENT
END:VCALENDAR