Question icon
Your Current Search
Choose below to refine your search
Research Topic
Download abstract book

Download the NI2012 abstract book here. The page numbers in the index are clickable for easy browsing.


From evolving artificial gene regulatory networks to evolving spiking neural networks for signal processing


Ahmed Abdelmotaleb (Adam Mickiewicz University), Borys Wróbel (Adam Mickiewicz University)

We extended the GReaNs platform (the name stands for Gene Regulatory evolutionary artificial Networks [1]) to enable for the evolution of spiking neural networks. The model of gene regulatory network used in the platform has been previously shown to be evolvable in tasks involving signal processing and animat control [1]. The structure of the network in this model is encoded in a linear genome, without imposing any restrictions on the size of the genome or the size of the network. In the previous work using GReaNs [1] each node in the artificial network has been considered to be an analog of a biological transcriptional unit. However, they could have equally well been seen as artificial neurons. Our current work goes further in the direction of artificial neural networks: we have introduced to GReaNs two models of spiking neurons (LIF: leaky integrate and fire neurons with a fixed threshold [2] and AdEx: adaptive-exponential integrate and fire [3]). The linear genome can now encode a network of these neurons and a genetic algorithm can be used to evolve networks with a particular spiking pattern, for example, an output of one spiking neuron to a specific input shifted by a specified time interval (Figure). We now intend to test the evolvability of the model in more demanding signal processing tasks.

[1] B. Wróbel and M. Joachimczak (2011) Using GReaNs (Genetic Regulatory evolving artificial Networks) for 3-dimensional Asymmetrical Pattern Formation and Morphogenesis. Proceedings of DevLeaNN: Workshop on Development and Learning in Artificial Neural Networks, pp. 26–28.
[2] P. Dayan and L. F. Abbott (2001) Theoretical neuroscience. Boston: MIT Press.
[3] W. Gerstner and R. Brette (2005) Adaptive exponential integrate-and-fire model as an effective description of neuronal activity. J. Neurophysiol. 94:3637-3642.
From evolving artificial gene regulatory networks to evolving spiking neural networks for signal processing
Figure. The membrane potential of the output neuron of a network of AdEx neurons evolved in GReaNs (blue line) to match the spikes of one LIF neuron (red line), shifted by 5 ms. Top panel: in response to the Poisson spike train (green) used during evolution. Bottom: a response to a different spike train, not used during evolution.
Preferred presentation format: Poster
Topic: Computational neuroscience