NEAT

Description

This repository contains an implementation of the NEAT algorithm (NeuroEvolution of Augmenting Topologies) from scratch. NEAT is a genetic algorithm for the generation of evolving artificial neural networks. The implementation is modular and flexible, allowing you to easily pass a custom fitness function to evolve neural networks tailored to your specific problem.

Features

• Modular Design: Easily integrate your own fitness function to customize the evolutionary process.
• NeuroEvolution: Evolve neural networks using genetic algorithms.
• Topological Innovation: Supports dynamic growth and complexity of neural networks through mutation operations that add nodes and connections.
• NumPy Powered: Efficient numerical computations using the NumPy library.

Getting Started

Prerequisites

• Python 3.x
• NumPy
• TensorFlow
• TQDM
• Matplotlib
• Gymnasium
• NetworkX

Installation

1. Clone the repository:

   git clone https://github.com/larrie1/neat.git
   cd neat
   

2. Install the required dependencies:

   pip install numpy tensorflow tqdm matplotlib gymnasium networkx
   

Usage

1. Define your fitness function:

   def fitness_function(genome):
    # Your custom fitness logic here
    return fitness_score
   

2. Initialize the NEAT algorithm:

   neat = NEAT(fitness_function=fitness_function, population_size=100, input_size=5, output_size=2)
   

3. Run the evolutionary process:

   best = neat.fit(generations=10, selection_size=20)
   

4. Plot the results of the best network:

   plot_genom(best, fitness)