/radboud-naco-project

Primary LanguageJupyter Notebook

Image Classification with Puissance-assisted Neuroevolution

"Image Classification with Puissance-assisted Neuroevolution" is a Natural computing project made by students of Radboud University.
This project applies Mutational Puissance Assisted Neuroevolution to NEAT.

Installation

The code on this repository was tested on Python 3.6.9.

Use the package manager pip to install the dependencies for this project from the root folder.

pip install -r requirements.txt

Optionally, you could create a virtual environment using the Python virtualenv package by running

virtualenv <env_name>
source <env_name>/bin/activate
pip install -r requirements.txt

Usage

There are 2 ways this project can be ran:

  • Run Sklearn-neuroevolution on the CIFAR-10 dataset.
  • Run Sklearn-neuroevolution with Mutational Puissance on the CIFAR-10 dataset.

To run the normal Sklearn-neuroevolution on the CIFAR-10 dataset.

mv ./Sklearn-neat/neat /Sklearn-neat/temp-neat
python ./Sklearn-neat/main.py

changing the directory name for neat will ensure that Sklearn-neuroevolution will import neat normally instead of using the locally modified version that should be used in th example below.

To run Sklearn-neuroevolution with Mutational Puissance on the CIFAR-10 dataset.

python ./Sklearn-neat/main.py

Contributors


Sebastiaan Ram
Ylja van Son
Ward Theunisse