|
Ultra fast cnn based Hardware Computing Platform Concepts for adas visual Sensors and Evolutionary Mobile RobotsBog'liq Alireza Fasih5.2.5
Selection theory in the genetic algorithms
As the process of evolution creates new generations and since every new generation
contains better chromosomes than the previous one, the process of selecting better
chromosomes for a next generation is very important. The selection process does not
consider the chromosomes merely on the basis of fitness. This process is rather random,
through which the parents for the new generations are selected randomly based on their
fitness. The method of selection used in our genetic algorithm approach is based on
selecting parents from half of the population. After evaluating the fitness of all the
chromosomes, the population is sorted in descending order. The best chromosomes
occupy higher locations in the list. Every time the two chromosomes are selected randomly
for crossover and mutation from half of the population. In order to increase the probability
of selecting good parents, the parents can be selected from a specified fraction of the half of
the population. For example, the first parent can be selected from half of the population
and the second can be selected from 30% of the half of the population from the pool of
much better parents.
5.2.6
Reproduction in the genetic algorithms
The reproduction phase involves generating a new population for the next generation from
the selected parents by applying two genetic operators, crossover and mutation on the
selected parents. These two processes result in the next generation population of
chromosomes that is different from the previous generation. In our approach, after
parents’ selection, crossover, mutation and fitness evaluation, the two children can go to
the next generation only when the fitness of each one of them is greater than the worst
child. The first child is compared with the worst parent and if its fitness is greater than the
worst parent, it is replaced by the worst parent and the population is sorted in descending
order. The same is done with the second child.
|
| |
