jcgp.backend.modules.problem

Class SymbolicRegressionProblem

java.lang.Object

jcgp.backend.modules.Module

jcgp.backend.modules.problem.Problem

jcgp.backend.modules.problem.TestCaseProblem<java.lang.Double>

jcgp.backend.modules.problem.SymbolicRegressionProblem

public class SymbolicRegressionProblem
extends TestCaseProblem<java.lang.Double>

Symbolic regression functions

Using this problem type, regression problems can be solved. parseData() must be used to load the desired function data in the standard CGP .dat format.

This problem uses quite a few parameters:

• Error threshold: the maximum difference allowed between an evolved output and the equivalent output from the problem data. Outputs within the error threshold will be considered correct. This is only used if hits is enabled.
• Perfection threshold: if the fitness is calculated without using the hits method, it is a decimal value. A solution is considered perfect when the difference between its fitness and the maximum possible fitness is within the perfection threshold.
• Hits-based fitness: increment the fitness by 1 whenever the chromosome output is within the error threshold.

Author:

Eduardo Pedroni

See Also:

SymbolicRegressionFunctions

Nested Class Summary

Nested classes/interfaces inherited from class jcgp.backend.modules.problem.TestCaseProblem

TestCaseProblem.TestCase<U>

Constructor Summary

Constructors

Constructor and Description
SymbolicRegressionProblem(Resources resources) Creates a new instance of SymbolicRegressionProblem.

Method Summary

Modifier and Type	Method and Description
void	evaluate(Population population, Resources resources) The most important method of the problem type.
int	hasImprovement(Population population) Used to assert whether a given population has a chromosome that is an improvement over the current best chromosome.
TestCaseProblem.TestCase<java.lang.Double>	parseTestCase(java.lang.String[] inputs, java.lang.String[] outputs) This method is used internally by addTestCase() in order to appropriately parse strings into the right data type for the test cases.
int	perfectSolutionFound(Population population) Used to assert whether a given population contains a perfect solution to the problem.

Methods inherited from class jcgp.backend.modules.problem.TestCaseProblem

addTestCase, clearTestCases, getTestCases, parseProblemData

Methods inherited from class jcgp.backend.modules.problem.Problem

getBestFitness, getFileExtension, getFitnessOrientation, getFunctionSet, reset

Methods inherited from class jcgp.backend.modules.Module

getLocalParameters, toString

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

SymbolicRegressionProblem

public SymbolicRegressionProblem(Resources resources)

Creates a new instance of SymbolicRegressionProblem.

Parameters:

resources - a reference to the experiment's resources.

Method Detail

evaluate

public void evaluate(Population population,
                     Resources resources)

Description copied from class: Problem

The most important method of the problem type. This is called once per generation, when the new population has been generated.

The basic functionality of this method is to loop through all chromosomes in the population and decode them according to the problem type. The fitness of each chromosome is then calculated using the problem data or otherwise (subjective problem types such as art generation might leave fitness evaluations up to the user) and assigned to the appropriate chromosome.

In addition, realisations of this method should update the value of bestFitness as appropriate, since the value of this parameter is displayed if a GUI is in use.

Specified by:

evaluate in class Problem

Parameters:

population - the population to be evaluated.

resources - parameters and utilities for optional reference.

parseTestCase

public TestCaseProblem.TestCase<java.lang.Double> parseTestCase(java.lang.String[] inputs,
                                                                java.lang.String[] outputs)

Description copied from class: TestCaseProblem

This method is used internally by addTestCase() in order to appropriately parse strings into the right data type for the test cases. Since the data type is problem-dependent, subclasses must implement this method. This method must return a built TestCase object from the arguments given.

Parameters:

inputs - the inputs represented as strings.

outputs - the outputs represented as strings.

Returns:

the parsed test case.

perfectSolutionFound

public int perfectSolutionFound(Population population)

Description copied from class: Problem

Used to assert whether a given population contains a perfect solution to the problem. It is up to the problem to define what qualifies a perfect solution, as some problems (subject ones such as music and art evolution, for example) might not have perfect solutions at all.

Specified by:

perfectSolutionFound in class Problem

Parameters:

population - the population to search through for a perfect chromosome.

Returns:

the perfect solution index, if one exits, -1 if no perfect solution was found.

hasImprovement

public int hasImprovement(Population population)

Description copied from class: Problem

Used to assert whether a given population has a chromosome that is an improvement over the current best chromosome. A typical implementation of this method will simply compare chromosome fitness values, though the problem type is free to implement this in any way.

Specified by:

hasImprovement in class Problem

Parameters:

population - the population potentially containing a fitter chromosome.

Returns:

the index of the first chromosome in the population that is an improvement, -1 if none is found.