Sunday 25 September 2016
Log in Register

Log in

The European Institute for Theoretical Neuroscience (EITN)

The European Institute for Theoretical Neuroscience (EITN) was created on March 2014 as part of the Theoretical Neuroscience activities of the Human Brain Project (HBP) and is operated by the UNIC unit of the CNRS.
The EITN has for scientific director Alain Destexhe (UNIC, CNRS, Gif-sur-Yvette) and is hosted by the Fondation Voir et Entendre & the Vision Institute in Paris.



We were happy to welcome Valentina Gliozzi (Università di Torino), Marco Brigham (Brigham Associates BVBA), Fabian Chersi (UCL), Morgan Taylor (University of Pennsylvania), Serge Korogod (ICMP) and Fabio Vallone (CNR) at the EITN to collaborate with our team


  • CDP5 kick-off meeting, May 12-13: “Functional plasticity for learning in large-scale systems”. 
    The workshop will focus on Product 1 of CDP5, "Suite of benchmark learning tasks for the 4 platforms (Heidelberg Physical Model, SpiNNaker & NEST & HPC, cortical column models, neurorobotics)”.
  • The Early Visual System : comprehensive and data-driven modelling, May 19-20 
    The purpose of this international workshop is to gather an expert group of 30-40 theoreticians and experimentalists in the field to take stock and debate about the current challenges when trying to understand the computations performed by the early visual system on the basis of multiscale data obtained by electrophysiological and calcium imaging data. More information here here
  • A SIMULATOR FOR POPULATION-LEVEL ACTIVITY HAS BEEN PRODUCED. A simulator – MIIND - for population-level activity has been produced, allowing the simulations of network of populations of spiking point model neurons. Any 1D neural model can be simulated. Large networks of populations can be simulated efficiently using MPI. Synaptic efficacies can be arbitrary large. SPs involved: SP4.
    In Task 4.4.1, we have implemented a model of dendritic excitability using the AdEx model, so fully compatible with neuromorphic hardware. This model is presently studied to reproduce dendritic spikes.

    Proof point/publication link here.The website will be updated shortly.
  • PUBLICATIONS: PROBABILISTIC MODELS. This research has re-examined the conceptual and mathematical framework for understanding the plasticity of networks of neurons in the light of experimental data that point to ongoing rewiring and drifts of neural codes even in the adult cortex. The resulting new theory proposes to view learning as probabilistic inference, more precisely as sampling of network configurations from a posterior distribution, It provides a principled understanding of conditions that enable stable learning and automatic compensation for perturbations.
    SPs involved : SP4, SP 9

    Network plasticity as Bayesian inference.
    D. Kappel, S. Habenschuss, R. Legenstein, and W. Maass.
    PLOS Computational Biology, in press, 2015.

    Synaptic sampling: A Bayesian approach to neural network plasticity and rewiring.
    D. Kappel, S. Habenschuss, R. Legenstein, and W. Maass.
    Proc. of NIPS 2015: Advances in Neural Information Processing Systems, in press, 2015.

  • PUBLICATIONS: MODEL OF ORCHESTRATED SYNAPTIC PLASTICITY. Models of Synaptic Plasticity have been consolidated and validated in a model of memory formation. The new model combines Hebbian spike-timing dependent plasticity with heterosynaptic plasticity. The model is compatible with experimental data on synapses and useful for the formation of new memories. The synaptic model gives ris to a stable algorithms that can be transferred from SP4 to various HBP Platforms, in particular the neuromorphic hardward.
    SPs involved: SP4

    Proof point/publication link here.