markdown

Neurogas

Status: 🚧 In development
Purpose: Learning project & portfolio showcase
Note: The API is experimental and subject to change.

Overview

Neurogas is a Rust implementation of Growing Neural Gas (GNG), an unsupervised learning algorithm for topology discovery and clustering.
This crate is primarily developed for educational purposes and as part of my portfolio. While it is functional, the API is evolving and may change significantly between versions.

Features

• 🧠 Growing Neural Gas implementation
• 🔄 Adaptive topology learning
• ✍️ Built as a demonstration of Rust and open-source contribution

Installation

Cargo.toml

[package]
name = "gng_test_app"
version = "0.1.0"
edition = "2021"

[dependencies]
neurogas = {version = "0.0.2"}

use::neurogas::Gng;

fn main() {
    const DATA_FILE: &str = "../blobs.csv";
    const OUTPUT_FILE: &str = "/tmp/output.json";
    let mut gng = Gng::new();

    gng.set_parameters(
        2,      // input_width
        -1.0,   // weight_rng_min
        1.0,    // weight_rng_max
        50,     // edge_removal_age
        200,    // neuron_creation_interval
        10000,  // max_train_iterations
        0.096,  // target_error
        0.1,    // epsilon_w
        0.006,  // epsilon_n
        0.5,    // alpha
        0.995,  // beta
    );

    gng.init_dataset(DATA_FILE);
    gng.fit();
    gng.save_model_json(OUTPUT_FILE);
}

Disclaimer

Not production-ready. Built for learning, experimentation, and portfolio use only.

This project is not maintained and may be abandoned permanently, especially if I move into a non-coding or non-IT role.

No support, no updates, no guarantees — fork it if you need it.