maael / spikingneuralnetwork

A python implementation of a spiking neural network with STDP and Reinforcement learning methods

Spiking Neural Network

This is a simple python implementation of a Spiking Neural Network (SNN) using spiking neuron models, with some Spike Timing Dependent Plasticity based learning methods.

Requirements

• python 3

Installation

pip3 install -r requirements.txt

Neuron Models

It includes two neuron models, the simplified Leaky Integrate and Fire model and the biophysically inspired Hodgkin-Huxley model.

Note - Only the Integrate and Fire neuron model is implemented fully.

Learning Methods

It includes three learning methods:

• Basic Spike Timing Dependent Plasticity (STDP)
• Homeostatic STDP variant
• Hedonistic reward-based Reinforcement Learning (RL) STDP variant

Usage

Include as a dependency

Add the following to your projects requirements.txt:

-e git+https://github.com/maael/SpikingNeuralNetwork.git#egg=SpikingNeuralNetwork

Import components

from snn.neurons.LeakyIntegrateAndFireNeuron import LeakyIntegrateAndFireNeuron
from snn.learning.stdp import STDP
from snn.network.snn import SNN

Create network

network = SNN(total_input_neurons, [hidden_neurons_1, hidden_neurons_2], total_output_neurons, LeakyIntegrateAndFireNeuron, STDP())

Tools

Visualisation Tool

This repository also includes a simple visualisation tool that is intended to allow viewing of the network structure and firing over time.

usage

python3 snn/tools/visualise.py

spikingneuralnetwork's Issues

about the reinforcement learning

where is the reinforcement learning and what is the principle is there any equation? how to use the reinforcement learning?