dpshorten / EC14-HyperNEAT

HyperNEAT

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

HyperNEAT EC class @VU 2014

Documentation on HyperNEAT

Usage:

cd into the out directory, and then:

  1. ./HyperNEAT -I <experiment>.dat -R $RANDOM -O <outputname>
    this will cause our version of the HyperNEAT tool to generate one individual with given output name. An individual consist of one .xml file (genotype) and .vxa file (phenotype)

  2. ./HyperNEAT -I <experiment>.dat -R $RANDOM -O <outputname> -ORG <individual01>
    HyperNeat tool will mutate the given individual and output this in the output name

  3. ./HyperNEAT -I <experiment>.dat -R $RANDOM -O <outputname> -ORG <individual01> <individual02>
    HyperNeat tool will generate a new individual with given outputname, based on the genetic code in individual01 and individual02

Explanation of the input arguments:

  • -I Denotes the input file with experiment instructions
  • -R The random seed
  • -O The output file (just the name of the individual, no file ending)
  • -ORG Existing additional input files (genotype only, again, only the name)

Output

Each command will generate 2 files in the (out) folder: [name]_genome.xml and [name]_vox.vxa, where [name] is the value of the -O argument. When you want to mutate or crossover, you only have to supply this [name], not the full [name]_genotype.xml to the -ORG argument.

Additional info:

This version of HyperNEAT will be used to research softrobot evolution without the use of an artificial fitness function. As such this version of HyperNEAT only supports reproduction of said softrobots.

$RANDOM is used so that it does not require to edit the parameter file each time and change the random seed value

future link to paper?

alt text

About

HyperNEAT


Languages

Language:C++ 75.3%Language:C 22.8%Language:Objective-C 1.6%Language:Shell 0.3%