This repository contains experiments in digital spiking neural networks.
The work is inspired by the Simple Model of Spiking Neurons by Eugene M. Izhikevich. However, we are attempting to build something that has a greater hardware-affinity while retaining the important properties of the spiking networks.
First, we want to build a spiking neural network that can play a simple game of tracking the dot with an eye:
Then, we want to make a network that can adapt (without expensive recomputation) even if the rules of the game change slightly.
Versions from one to 5 attempt to come up with a model that is computationally simple (e.g. doesn't have multiplication or division) but can still exhibit complex behavior.
In V6 we rewrote the core code in golang to speed up the computation and run larger networks. We are still manually tunning the parameters and the network model here.
