Fuzzy Clustering and Genetic Algorithms - Biomedical Signal Analysis: Contemporary Methods and Applications - page 242

These important and most common operators are the following [282]:

• Encoding scheme:

Transforms pattern vectors into bit string representations. Each coordi-
nate value of a feature vector can be encoded as a binary string. Through
an ecient encoding scheme, problem-specific knowledge is translated
directly into the GA framework and implicitly influences the GA's per-
formance.
•

Fitness evaluation:
After the creation of a generation, fitness evaluation becomes important
in order to provide the correct ranking information necessary for per-
petuation. Usually, fitness of a member is related to the evaluation of
the objective function of the point representing this member.
• Selection:

Based on selection , population members are chosen based on their fitness
(the value of the objective function for that solution). The strings in
the current population are copied in proportion to their fitness and
placed in an intermediate generation. Selection enables the fittest genes
to perpetuate, and guarantees the convergence of the population toward
the desired solution.
•

Crossover:
Crossover describes the swapping of fragments between two binary
strings at a random position and combines the head of one with the
tail of the other, and vice versa. Thus, two new offsprings are created
and are inserted into the next population. In summary, new sample
points are generated by recombining two parent strings. Consider the
two strings 000101000 and 111010111. Using a single randomly chosen
crossover point, recombination occurs as follows:

000 | 101000

111 | 010111 .

The following offsprings are produced by swapping the fragments be-
tween the two parents:

000010111 and 111101000

This operator also guarantees the convergence of the population.