synapticflow 0.0.4

A powerful Python package for simulating spiking neural networks (SNNs) using PyTorch with GPU acceleration.

SynapticFlow

Spiking Neural Networks (SNNs) are a type of artificial neural network that attempts to mimic the behavior of neurons in the brain. Unlike traditional neural networks that use continuous-valued signals, SNNs operate using discrete spikes of activity that are similar to the action potentials in biological neurons. SynapticFlow is a powerful Python package for prototyping and simulating SNNs. It is based on PyTorch and supports both CPU and GPU computation. SynapticFlow extends the capabilities of PyTorch and enables us to take advantage of using spiking neurons. Additionally, it offers different variations of synaptic plasticity as well as delay learning for SNNs.

Please consider supporting the SynapticFlow project by giving it a star ⭐️ on Github, as it is a simple and effective way to show your appreciation and help the project gain more visibility.

If you encounter any problems, want to share your thoughts or have any questions related to training spiking neural networks, we welcome you to open an issue, start a discussion, or join our Discord channel where we can chat and offer advice.

Installation

To install synapticflow, run the following command in your terminal:

$ pip install synapticflow

We recommend using this method to install synapticflow since it will ensure that you have the latest stable version installed.

If you prefer to install synapticflow from source instead, follow these instructions:

$ git clone https://github.com/arsham-khoee/synapticflow
$ cd synapticflow
$ python setup.py install

Requirements

The requirements for SynapticFlow are as follows:

• torch
• matplotlib

Usage

After package installation has been finished, you can use it by following command:

import synapticflow as sf

In following code, a simple LIF neuron has been instantiated:

model = sf.LIFPopulation(n=1)
print(model.v) # Membrane Potential
print(model.s) # Spike Trace

SynapticFlow Structure

The following are the components included in SynapticFlow:

Component	Description
synapticflow.network	A spiking network components like neurons and connections
synapticflow.encoding	Several encoders implementation
synapticflow.learning	Learning rules and surrogate gradients
synapticflow.evaluation	Several evaluation functions for networks
synapticflow.datasets	Include MNIST, Fashion-MNIST, CIFAR-10 benchmark datasets
synapticflow.vision	Include vision components for neuroscience
synapticflow.plot	Plot tools for neural networks visualization