Custom Selection Scheme

This tutorial introduces the use of custom selection schemes.

By default the CGP-Library uses the select fittest selection schemes which simply selects the mu best members of the candidate chromosomes to become the parents.  This tutorial will demonstrate how to define and use alternative custom selection schemes.

A simple program showing how to uses a custom tournament selection scheme.

The program below is provided in the CGP-Library download within /examples/customSelectionScheme.c and is included in the code::blocks project.

#include <stdio.h>
#include <stdlib.h>
#include "../src/cgp.h"

void tournament(struct parameters *params, struct chromosome **parents, struct chromosome **candidateChromos, int numParents, int numCandidateChromos){

    int i;

    struct chromosome *candidateA;
    struct chromosome *candidateB;

    for(i=0; i<numParents; i++){

        candidateA = candidateChromos[rand() % numCandidateChromos];
        candidateB = candidateChromos[rand() % numCandidateChromos];

        if(getChromosomeFitness(candidateA) <= getChromosomeFitness(candidateB)){
            copyChromosome(parents[i], candidateA);
        }
        else{
            copyChromosome(parents[i], candidateB);
        }
    }
}

int main(void){

    struct parameters *params = NULL;
    struct dataSet *trainingData = NULL;
    struct chromosome *chromo = NULL;

    int numInputs = 1;
    int numNodes = 50;
    int numOutputs = 1;
    int arity = 2;

    double targetFitness = 0.1;
    int updateFrequency = 1000;
    int numGens = 10000;

    params = initialiseParameters(numInputs, numNodes, numOutputs, arity);

    addNodeFunction(params, "add,sub,mul,div,sin");

    setTargetFitness(params, targetFitness);

    setUpdateFrequency(params, updateFrequency);

    setCustomSelectionScheme(params, tournament, "tournament");

    trainingData = initialiseDataSetFromFile("./dataSets/symbolic.data");

    chromo = runCGP(params, trainingData, numGens);

    freeChromosome(chromo);
    freeDataSet(trainingData);
    freeParameters(params);

    return 0;
}

A close look at each line of the example code.

First standard headers and cgp.h are included.

#include <stdio.h>
#include <stdlib.h>
#include "../src/cgp.h"

Next we defined our custom selection scheme to be used.

The custom selection scheme prototype must take the following form.  Where params is a pointer to an initialised parameters structure, parents is a pointer to an array of chromosomes which will contain the parents when the selection scheme terminates, candidateChromos is a pointer to an array of chromosomes which contains the chromosomes from which the parents will be selected, numParents give the number of parent chromosomes to be selected and numCandidateChromos gives the number of candidate chromosomes to select from.

void selectionSchemeName(struct parameters *params, struct chromosome **parents, struct chromosome **candidateChromos, int numParents, int numCandidateChromos);

Here a tournament selection scheme is defined which uses a tournament size of two. candidateA and candidateB are used to point to two randomly selected chromosomes from the candidate chromosomes.  The fitter of the two randomly selected chromosomes is then copied into the array of parent chromosomes using copyChromosome.

void tournament(struct parameters *params, struct chromosome **parents, struct chromosome **candidateChromos, int numParents, int numCandidateChromos){

    int i;

    struct chromosome *candidateA;
    struct chromosome *candidateB;

    for(i=0; i<numParents; i++){

        candidateA = candidateChromos[rand() % numCandidateChromos];
        candidateB = candidateChromos[rand() % numCandidateChromos];

        if(getChromosomeFitness(candidateA) <= getChromosomeFitness(candidateB)){
            copyChromosome(parents[i], candidateA);
        }
        else{
            copyChromosome(parents[i], candidateB);
        }
    }
}

Much of the main function has been introduced in the Getting Started tutorial and is not re-described here.

In order to use our new tournament selection scheme it must be set in the parameters structure using setCustomSelectionScheme.  Where the first variable is a parameters structure, the second is our custom selection scheme and the third is a user defined name for the selection scheme.

Now when runCGP or repeatCGP is called the this custom selection scheme will be used in place of the default select fittest.

setCustomSelectionScheme(params, tournament, "tournament");

And that’s it, in order to use a custom selection scheme it must be defined using the specific prototype and then added as the selection scheme to be used using by calling setCustomSelectionScheme.  To revert back to the default selection scheme simply call setCustomSelectionScheme with the selection scheme variable set as NULL.