Spiking Neural Network - A Rust Implementation ⚡️🦀⚡️

We consider (continuous-time) spiking neural networks and their use as robust memorizers of spike scores. We refer to Aguettaz and Loeliger, 2024 for details.

The implementation is based on the Rust programming language. This language was designed with several key goals, which revolve around improving safety, performance, and concurrency.¹

Installation

To install Rust, follow the instructions on the official website. The installation includes the Rust compiler, Cargo (the Rust package manager), and the Rust standard library.

Building From Source

To build the project from source, clone the repository and execute the following command in the root directory:

cargo build --release

You can generate and view the rusty-snn documentation locally by running:

cargo doc --open

Installation From Crates.io

This feature will be available soon.

Usage

An example of how to use the library is provided in the examples directory.

• example1.rs demonstrates how to create a network randomly, train it to memorize a random spike train, and simulate the network.

References

• Aguettaz and Loeliger, "Continuous-time neural networks can stably memorize random spike trains", arXiv, 2024.

• Klabnik and Nichols, The Rust Programming Language, 2021.

Footnotes

1. Read as concurrency or parallelism. ↩