SENSOPAC
Structure de mise en forme 2 colonnes
  • Thursday 19 November 2015
  • EDLUT: spiking network simulation

    Event-Driven simulator based on Look-Up-Tables

    The EDLUT open source spiking neural simulator has been developed in the framework of SENSOPAC, the code can be found here. Therefore, now the whole computational neuroscience research community can freely access to the simulator and further develop it according to its new necessities. See more details, a hello world example and an explanatory presentation at the link above.

    LWPR: Locally Weighted Projection Regression

    Locally Weighted Projection Regression

    Locally Weighted Projection Regression (LWPR) is a recent algorithm that achieves nonlinear function approximation in high dimensional spaces with redundant and irrelevant input dimensions. At its core, it uses locally linear models, spanned by a small number of univariate regressions in selected directions in input space. A locally weighted variant of Partial Least Squares (PLS) is employed for doing the dimensionality reduction.

    Context Estimation and Switching

    We investigate how a robot can possibly discriminate between objects through interaction. To this end, we develop algorithms for estimating and switching between contexts (e.g. different loads, moments of inertia, texture). Our work can be roughly divided into the three cases of

    • discrete contexts,
    • continuous contexts, and
    • discrete/continuous context with additional sensory information.

    Software and data sets for these experiments can be found here.

    Granular layer network model

    The way the cerebellar granular layer transforms incoming mossy fiber signals into new spike patterns to be related to Purkinje cells is not yet clear. Here, a realistic computational model of the granular layer was developed and used to address four main functional hypotheses: center-surround organization, time-windowing, high-pass filtering in responses to spike bursts and coherent oscillations in response to diffuse random activity. The model network was activated using patterns inspired by those recorded in vivo. Burst stimulation of a small mossy fiber bundle resulted in granule cell bursts delimited in time (time windowing) and space (center-surround) by network inhibition. This burst-burst transmission showed marked frequency-dependence configuring a high-pass filter with cut-off frequency around 100 Hz. The contrast between center and surround properties was regulated by the excitatory-inhibitory balance. The stronger excitation made the center more responsive to 10-50 Hz input frequencies and enhanced the granule cell output (with spikes occurring earlier and with higher frequency and number) compared to the surround. Finally, over a certain level of mossy fiber background activity, the circuit generated coherent oscillations in the theta-frequency band. All these processes were fine-tuned by NMDA and GABA-A receptor activation and neurotransmitter vesicle cycling in the cerebellar glomeruli. This model shows that available knowledge on cellular mechanisms is sufficient to unify the main functional hypotheses on the cerebellum granular layer and suggests that this network can behave as an adaptable spatio-temporal filter coordinated by theta-frequency oscillations.

     

    Requirements and Instructions

    The network model (download) was built to run on the NEURON 7.2 simulation program. The network model was tested to run also under the previous NERUON versions till the 6.0 under Windows, Linux and Mac OS X. To run the network model access the model directory, compile the mod files accordingly to the standard procedure required by the chosen operative system (instructions can be found at www.neuron.yale.edu/neuron), and run the Start.hoc script. All the simulation data will be stored in the SimData directory in ASCII code files. The actual model configuration is designed to store all mossy fiber, Golgi cell, and granule cell spike trains and the granule and Golgi cell membrane potential traces. In MATLAB from the matlab_view directory run

    >> simdata = Vis_net('../SimData')

    to load the simulation data into MATLAB workspace and plot color coded views of the activity of granule cells in the network. The script will save a SimData.avi file in the SimData directory.

    The simulation setup is now configured to reproduce the activity of a cube of 100 \xB5m edge length of the granular layer with 4000 granule cells, 23 Golgi cells, and 320 mossy fiber terminals. The simulation covers about 400 ms and a burst of 3 spikes at 300 Hz is delivered at 250 ms from simulation beginning by mossy fiber terminals located within a sphere (20 \xB5m radius) located in the center of the cube.

    UPMC Spike Train

    1. For the package "UPMC.Metrical.Information.Matlab.zip":

    UPMC - Metrical Information Analysis : Matlab functions for calculating the metrical information (Brasselet et al. 2011) using labelled output spike trains. This version uses the Victor and Purpura distance as a metrics for quantifying differences between spike trains. The metrical information theory was developed during the SENSOPAC project in order to analyse the efficiency of neural encoding/decoding mechanism underlying transmission of tactile information from peripheral mechanoreceptors to central nervous areas.

    2. For the package "UPMC.SpikeBuilder.zip"

    UPMC - SpikeBuilder C++ code : Program using inputs sent by artificial touch sensors (real or simulated) to generate spike train responses emulating mechanoreceptor and cuneate neuron activities. Output spikes are then sent to the Metrical Information Analysis Tool (running under Matlab). This software was used in a Braille reading task where information from a tactile sensor mounted on the DLR robotic hand were converted online into spike patterns and discriminated in real time by a naive bayesian classifier.