Implementation of Spiking Neural Networks

I’m working with the non-profit research Astera Institute on the Leabra framework, which uses a spiking model of neurons similarly to Nengo. We’re running into performance difficulties, and we’re thinking about using Nengo on the backend. I’d like to understand Nengo’s implementation a bit better so we can have a clearer idea of whether it would work for our use case.

Would anyone who’s more experienced with Nengo be willing to have a brief chat with me about the implementation? I’d also be curious to hear your thoughts about the theory underlying spiking nets, because I suspect there are computational improvements that come from a spiking representation over and above any efficiency gains.


Hi @qemqemqem, and welcome to the Nengo forums. :smiley:

Also, thank you for your interest in Nengo. We’d love to arrange a call with you to discuss some form of future collaboration. To discuss possible collaboration, you can email our CEOs, Chris ( and Pete ( Chris heads up the tech team, and Pete is in charge of the business related stuff.

If you have any basic questions regarding Nengo’s various implementations, I can reply to your questions here (on the forums), and I will forward your questions to our devs if they are beyond the scope of my knowledge.

As for thoughts about spiking networks, I’ll refer you to Chris. He’s the inventor of the NEF (Neural Engineering Framework) algorithm that makes it possible for us to efficiently design functional neural networks in spiking neurons; so he’s probably the best qualified to discuss that topic with you. :smiley:

1 Like