Decoding spiking motifs using neurons with heterosynaptic delays

Abstract

The response of a biological neuron depends largely on the precise timing of presynaptic spikes that reach the basal dendritic tree. However, most neuronal models do not take advantage of this minute temporal dimension, especially in exploiting the variety of synaptic delays on the dendritic tree. A notable exception is the polychronization model, a recurrent model of spiking neurons including fixed and random heterosynaptic delays and in which the weights are learned using Spike-Time Dependent Plasticity. The output raster plot displays repeated activations of prototypical spiking motifs called Polychronous Groups. Importantly, these motifs seem to be highly relevant in experimental neuroscience. Here, by extending the model of~[3], we develop a spiking neural network model for the efficient detection of PGs: By defining the generation of the raster plot as a probabilistic combination of PGs, we build and train the network in order to optimize the inversion of this generative model.

Publication
Proceedings of AREADNE
Antoine Grimaldi
Antoine Grimaldi
Phd candidate in Computational Neuroscience

During my PhD, I am focusing on Ultra-fast vision using Spiking Neural Networks.

Hugo Ladret
Hugo Ladret
Phd candidate in Computational Neuroscience

During my PhD, I am focusing on the role of precision in natural and artificial neural networks.

Laurent U Perrinet
Laurent U Perrinet
Researcher in Computational Neuroscience

My research interests include Machine Learning and computational neuroscience applied to Vision.