You are on page 1of 2

Genetic Algorithm (GA) is part of the Evolutionary Algorithm, an algorithm that imitates the

natural evolutionary process where the central concept is that the most superior individuals
will survive. In contrast, the weaker individuals will become extinct[1]. The advantages of
these individuals are tested through a function known as the fitness function. Fitness in GA is
defined as a description of the feasibility of a solution to a problem. The Fitness Function will
generate a fitness value that will be a reference for the following GA process. In general, the
GA process is shown in Figure 1

The GA process begins by determining the initial population, consisting of several


chromosomes composed of several genes representing the candidate solutions to a problem.
The best candidates will be selected through a selection process based on the calculated
fitness value for each chromosome in the population. Candidates selected from this process
will fill the mating pool, which is a set where two parents will be formed from here. In the
Evolutionary Algorithm, the principle of survival arises because of the reproduction process.
The resulting offspring will carry the traits of their parent's genes (parents); therefore, parents
are selected from the mating pool, a collection of the best candidates from a population. Thus
the resulting derivative is a derivative that has superior properties from both parents.

GA is one of the algorithms used for the optimization process. In optimization, the optimal
condition of the obtained solutions is the main target to be achieved. However, in the
optimization algorithm, local optimum conditions often occur. The local optimum is when the
algorithm reaches the highest or lowest value on some candidate solution values. This is in
contrast to the global optimum condition, where the algorithm gets the highest or lowest
value for all candidate solutions in a particular problem. One of the causes of local optimality
is the population reaching the convergence format too early, premature convergence.
According to Rajeev and Krishnamoorthy [2], the criteria for achieving convergence is when
about 80% or 85% of the number of chromosomes have the same gene values. One way to
prevent this premature problem is to maintain the chromosomal diversity of a population. In
GA, the chromosomal diversity of a population can be supported by implementing crossover
and mutation operators.

Crossover is a recombination operator that aims to get a better individual. The crossover
operator recombines the parent set, which will be chosen randomly from the mating pool
formed from the selection process. The crossover will produce a set of offspring whose
diversity will be maintained with the following procedure, namely mutation. In the mutation
operator, variety will be held by exchanging genes in the chromosome with the opposite
value. For example, if our chromosomes have binary values of 0 and 1, then if a randomly
selected mutation point has a value of 1, this value will be exchanged for a value of 0 or vice
versa. The result of this mutation operator is a new derivative which will then be re-tested on
the fitness function to see the feasibility of the new population from the results of this GA
process as a candidate solution to the given problem. The process of testing fitness, selection,
crossover and mutation will be repeated so that one of the following GA process control
loops has been fulfilled, namely iteration, convergence or fitness value.

You might also like