I am just starting a project using the Izhikevich model for a spiking nerual network and was wondering if anyone has any advise or help on how to get started with creating this?
what degree of details would you like to reach in your cuda application?
would you like to use conductance delay? Axonal/dendrites growth?
Basing on this you can start the organization of your model.
So, what would you like to do?
Personally I’m trying to implement all the known features using the CUDA.
Hope this gives you a direction …
Thanks your your reply, I’m not sure what I should include in my model, are you saying I need to plan out exactly what I’m going to be implementing before I actually start to do anything?
I think I would look at example of code dealing with particles and see how to
replace the particle behavior with the Izhikevich equations … Should not be
too much of an issue. Look also for some solving code PDE in financial apps.
I have implemented a first version of the model that incorporates
detailed neuron dynamics, STDP parameters, Axonol delay.
I am planning to release a code in a few days and would post the link in this forum.
Hope you can get more ideas or inspiration from that version.
Details of the code is available in the following paper…
Please let me know if you have any questions or comments…
gpusnn.pdf (329 KB)