/DeepHyperNEAT-Julia

A unofficial julia implementation of the DeepHyperNEAT system for evolving neural networks. Written in julia by Gabriel Diaz, developed by Felix Sosa and Kenneth Stanley (all credit goes to them and all blame to me). See paper: https://eplex.cs.ucf.edu/papers/sosa_ugrad_report18.pdf and original code: https://github.com/flxsosa/DeepHyperNEAT

Primary LanguageJuliaApache License 2.0Apache-2.0

Deep HyperNEAT: Extending HyperNEAT to Evolve the Architecture and Depth of Deep Networks

This implementation is a julia translation of Felix Sosa's python code in https://github.com/flxsosa/DeepHyperNEAT, which itself takes some things from PurePLES python code in https://github.com/ukuleleplayer/pureples

The code was tested with julia 1.7, it migth work with julia 1.6 but not less.

The "main" branch code is a copy of Felix Sosa's code. We recommend to use the branch "start_modified" since we made some improvements in the code there.

Example

An example is provided in the file xor.jl. You can run it the following

julia> using Revise

julia> includet("xor.jl")

i.e., it assumes that you have instald Revise

Notes

The LICENSE file is a copy of Felix's LICENSE.

If you have any observation please let me know.