[Neurodebian-users] Possibilities of Building "Stacked Neural Networks"

[Yury V. Zaytsev]

Hi,

On Tue, 2013-09-24 at 14:12 -0400, Michael Lamport Commons wrote:
> 
>        Would it be possible to build “stacked neural networks” like
> the one shown in the attached document?

It is certainly possible to build such things with currently existing
software, but it will require larger or smaller amounts of custom
coding, depending on the chosen platform.

In particular, you didn't specify the degree of realism you want when
modelling the boxes (populations, point neurons, neurons with
complicated morphology, etc.), the details on how the connectivity is
organized, etc.

>From the rest of the document, I would assume that you are interested in
point neurons or populations. There are simulators that work on the
NeuroDebian platform which you can use to build such networks, like
NEST, or Brian, or NEURON. I'm less familiar with population level
software, but if memory serves me well, Nengo might fit the bill.

I think that you might get better response if you reach out to more
specialized communities; this list mainly deals with software issues.

-- 
Sincerely yours,
Yury V. Zaytsev