next up previous contents index
Next: Selection by deletion of Up: Evaluator for individuals Previous: Evaluator for individuals   Contents   Index


The fitness of individuals

The fitness of an individuals is determined by different fitness factors.

One of the most important is, to what extent the individual is similar to the desired original multimedia data (such as the original image). The more the individual (or the phenotype of it) is similar to the original multimedia data, the higher the fitness should be and the lower is the error, which the individual has for the display of the original multimedia data.

This error (and thus the fitness of the individual) can be determine for example, with the sum of (the squared) differences (not defined points give the maximum error) of the colors of the points between the original image and the image produced by the individual, or through another self-determined distance.

If the fitness of separate part objects of the individual should be determined, this can be achieved, for example, by including in the evaluation only the area that is covered by the part object, the covered area and a border around this area or only the smallest square, which encompass this part object.

Another useful fitness factor is the size (increases with the number of elements) of the separate individuals, to give bigger individuals a lower fitness than smaller individuals, with the same error on the original multimedia data, and to prefer the smaller individuals.

A further fitness factor, which can be included, is the estimate of the time it takes to evaluate the individual for displaying the multimedia object (phenotype). In this way maybe even the execution speed of the algorithm can be increased.


next up previous contents index
Next: Selection by deletion of Up: Evaluator for individuals Previous: Evaluator for individuals   Contents   Index
Betti Österholz 2013-02-13