- Poster presentation
- Open Access
Evolutionary algorithm search for network connectivities conducive to periodic behavior at sub-spiking frequencies
BMC Neuroscience volume 14, Article number: P393 (2013)
We use an evolutionary algorithm (EA) to search the space of leaky integrate-and-fire (LIF) neuron networks, in order to identify network connectivities producing significant rhythmic activity at sub-spiking frequencies (i.e., 'bursting-like behavior'). We find that the connectivities of the most-fit LIF networks exhibit a relatively broad in-degree distribution and a relatively narrow out-degree distribution. We examine the frequencies of connection motifs in the most-fit LIF networks as compared to random networks. In a network of more realistically modeled neurons, the most-fit network connectivities are observed to produce a broader frequency response as compared to that resulting from random network connectivities.
Figure 1 shows the cumulative in- and out-degree distributions of a representative most-fit network, whose connectivity graph is illustrated in the inset. The orange line represents the cumulative in-degree distribution, and the red line the cumulative out-degree distribution. The blue bars represent the expected cumulative distribution (and error bars) for random network connectivities.
About this article
Cite this article
Robb, D.T., Toporikova, N. Evolutionary algorithm search for network connectivities conducive to periodic behavior at sub-spiking frequencies. BMC Neurosci 14 (Suppl 1), P393 (2013). https://doi.org/10.1186/1471-2202-14-S1-P393
- Animal Model
- Frequency Response
- Algorithm Search
- Evolutionary Algorithm
- Cumulative Distribution