Class Individual

java.lang.Object
org.episteme.core.mathematics.optimization.evolutionary.geneticprogramming.Individual
public class Individual extends Object

This class combines a program and its computed fitness values.

Raw fitness is the measurement of fitness that is stated in the natural terminology of the problem itself. The most common definition of raw fitness is error: If the S-expression is boolean-valued or symbolic-valued, the sum of distances is equivalent to the number of mismatches. If the S-expression is real-valued or integer-valued, the square root of the sum of the squares of the distances can, alternatively, be used to measure the fitness. (therefore increasing the influence of more distant points)

Standardized fitness restates the raw fitness so that a lower numerical value is always a better value.

It is convenient and desirable to make the best value of standardized fitness equal 0. If this is not already the case, it can be made so by subtracting (or adding a constant).

If for a particular problem, a greater value of raw fitness is better, standardized fitness must be computed from raw fitness. In that situation, standardized fitness equals the maximum possible value of raw fitness(Rmax) minus the observed raw fitness. For ex. If the artifical ant finds 5 0f 89 pieces of food using a given computer program, the raw fitness is 5 and the standardized fitness is 84.

Adjusted fitness a(i,t) is computed from the standardized fitness s(i,t) as follows: --- a(i,t) = 1 / 1+s(i,t) --- where s(i,t) is the standardized fitness for individual i at time t.

The adjusted fitness lies between 0 and 1. The adjusted fitness is bigger for better individuals in the population. Note that for certain methods of selection other than fitness proportionate selection (e.g. tournament selection and rank selection), adjusted fitness is not relevant and not used.

Normalized fitness is also needed if the method of selection employed is fitness proportionate.

The normalized fitness n(i,t) is computed from the adjusted fitness a(i,t) as described in Koza's book (ISBN:0262111705)

Note that for certain methods of selection other than fitness proportionate selection (e.g. tournament selection and rank selection), normalized fitness is not relevant and not used.

Version:
0.1
Author:
Levent Bayindir

  • Constructor Details

    • Individual

      public Individual(Program program)
      Create an individual with given parameters.
      Parameters:
      program - program of this individual

  • Method Details

    • copy

      public Individual copy()
      Create a deep copy of this individual
      Returns:
      deep copy of the individual

    • getProgram

      public Program getProgram()
      Returns program of this individual
      Returns:
      program of this individual

    • setProgram

      public void setProgram(Program program)
      Changes the program of this individual
      Parameters:
      program - new program of this individual

    • getNormalizedFitness

      public double getNormalizedFitness()
      Returns normalized fitness of this individual
      Returns:
      normalized fitness of this individual

    • getAdjustedFitness

      public double getAdjustedFitness()
      Returns adjusted fitness of this individual
      Returns:
      adjusted fitness of this individual

    • getStandardizedFitness

      public double getStandardizedFitness()
      Returns standardized fitness of this individual
      Returns:
      standardized fitness of this individual

    • setNormalizedFitness

      public void setNormalizedFitness(double newFitness)
      Changes normalized fitness of this individual
      Parameters:
      newFitness - new fitness value

    • setAdjustedFitness

      public void setAdjustedFitness(double newFitness)
      Changes adjusted fitness of this individual
      Parameters:
      newFitness - new fitness value

    • setStandardizedFitness

      public void setStandardizedFitness(double newFitness)
      Changes standardized fitness of this individual
      Parameters:
      newFitness - new fitness value

    • setHits

      public void setHits(int newHits)
      Changes hits of this individual
      Parameters:
      newHits - new hits