https://github.com/n3pdf/evolutionary_keras

An evolutionary algorithm implementation for Keras
https://github.com/n3pdf/evolutionary_keras

evolutionary-algorithm genetic-algorithm keras machine-learning open-source python tensorflow

Last synced: 3 months ago
JSON representation

An evolutionary algorithm implementation for Keras

• Host: GitHub
• URL: https://github.com/n3pdf/evolutionary_keras
• Owner: N3PDF
• License: gpl-3.0
• Created: 2019-12-18T14:51:40.000Z (about 5 years ago)
• Default Branch: master
• Last Pushed: 2021-01-04T10:24:16.000Z (about 4 years ago)
• Last Synced: 2024-10-09T12:32:36.521Z (4 months ago)
• Topics: evolutionary-algorithm, genetic-algorithm, keras, machine-learning, open-source, python, tensorflow
• Language: Python
• Homepage:
• Size: 162 KB
• Stars: 10
• Watchers: 4
• Forks: 7
• Open Issues: 10
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.3630339.svg)](https://doi.org/10.5281/zenodo.3630339)

[![](https://github.com/N3PDF/evolutionary_keras/workflows/pytest/badge.svg)](https://pypi.org/project/evolutionary-keras/)

[![Documentation Status](https://readthedocs.org/projects/evolutionary-keras/badge/?version=latest)](https://evolutionary-keras.readthedocs.io/en/latest/?badge=latest)

[![Anaconda-Server Badge](https://anaconda.org/conda-forge/evolutionary_keras/badges/installer/conda.svg)](https://anaconda.org/conda-forge/evolutionary_keras)

# evolutionary_keras

Keras is one of the most widely used Machine Learning frameworks available in the market. It is a high-level API written in Python and that can run on mulitple backends. Their goal is to be able to build and test new model as fast as possible.

Keras models are trained through the usage of optimizers, all of which are Gradient Descent based. This module deals with that shortcoming of Keras by implementing several Evolutionary Algorithms on top of Keras while keeping the main philosophy of the project: it must be easy to prototype.

The default project library now provides support for:

- Nodal Genetical Algorithm (NGA)

- Covariance Matrix Adaptation Evolution Strategy (CMA-ES)

## Installation

evolutionary_keras is available to install from pip and conda using the following commands:

```bash

    pip install evolutionary-keras

    conda install evolutionary_keras -c conda-forge

```

After which it can readily be imported in your python project

```python

    import evolutionary_keras

```