0% found this document useful (0 votes)
15 views16 pages

ZOT L6GeneticAlgorithms - Tr.en

The document summarizes key aspects of genetic algorithms. It defines genetic algorithms as population-based algorithms that use techniques like selection, crossover and mutation to evolve solutions to problems iteratively. It describes the basic components of genetic algorithms like chromosomes, genes, populations, fitness functions. It then explains the main operations in genetic algorithms like initializing the population, evaluating individuals using fitness functions, selecting individuals for mating, performing crossover and mutation to create new offspring and generations.

Uploaded by

mouna
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
15 views16 pages

ZOT L6GeneticAlgorithms - Tr.en

The document summarizes key aspects of genetic algorithms. It defines genetic algorithms as population-based algorithms that use techniques like selection, crossover and mutation to evolve solutions to problems iteratively. It describes the basic components of genetic algorithms like chromosomes, genes, populations, fitness functions. It then explains the main operations in genetic algorithms like initializing the population, evaluating individuals using fitness functions, selecting individuals for mating, performing crossover and mutation to create new offspring and generations.

Uploaded by

mouna
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 16

Translated from Turkish to English - www.onlinedoctranslator.

com

Intelligent Optimization Techniques


Intelligent Optimization Techniques

Prepared by: M. Ali Akcayol


Gazi University
Department of Computer Engineering

Presentations for this course, “Singiresu S. Rao, Engineering Optimization: Theory and Practice, Wiley, 2009.”
Prepared using the book.

Contents

-genetic algorithm
-Genetic algorithm operations
-Creating the new generation
-algorithm

-Performance

2
genetic algorithm
-It was developed by John Holland in the 1970s.
-David E. Goldberg in 1989Genetic Algorithm Applications
Considered a classic work onpublished his book.

-It is a population based algorithm.


-Every individual is a candidate to be the solution to the problem.

-stochasticis an algorithm andIt randomly selects in multiple


stages.
-Solutions are encoded as bit strings or differently depending on
the problem.
-individualsevaluation function(fitness function)
problem orientedis the most important part.
-When the algorithm ends, the best individual in the population is
taken as the solution.
3

genetic algorithm
Definitions
-population
-constituting the solution setall individuals.
-In every generationpopulation size remains constant.

-Chromosome

-used for solutionare individuals.


-Eachchromosome solutionentirelyexpresses.

-Gene
-inside the chromosomeIt is the smallest meaningful information.

-in the problemIt is related to the number of parameters.


-in a chromosomeThe combination of all genes represents the
solution.

4
genetic algorithm
Definitions
-locus
-inside the chromosomeis the location where the gene is located.
-In some problems, there may be information assigned to each location (day,
time, etc.).

genetic algorithm
Definitions
-Mutation
-in a chromosomeIt is a random change.
-in the populationincreases diversity.
-Bychanging adaptivelyFalling into local minima can be
prevented.
-Generation
-It is a new community.
-population in each generationAs many new individuals are created
as there are.
-mating
-between individualsis matching.
-In every matchTwo individuals are used.
-between two individualsfeature transfer is made.
6
Contents

-genetic algorithm
-Genetic algorithm operations
-Creating the new generation
-algorithm

-Performance

Genetic algorithm operations

-solution orrepresentation of the individual.

-Creation of the initial population.

-Calculating the solution suitability value of individuals


(fitness function is used.)

-of individualsselection for new population

-NewOnecreation of generation

8
Genetic algorithm operations
Illustration of the solution or individual
-Individualone in the populationstands for chromosome.
-Individuals contain the entire solution to the problem.
-Individuals can be encoded as binary, real number, integer, character or
otherwise.

Genetic algorithm operations


Create the starting population
-A known set of solutionscan be started by taking it.
-random individualscan be created.
-Individuals who meet certain criteriacan be created.
-in the populationnumber of individualsGenerallyBetween 100 and 300 is
selected.

10
Genetic algorithm operations
Create the starting population
-Startingfrom the initial population, guidance is made towards the
global best solution.
-Every individual,in solution spacerepresents a candidate solution.

11th

Genetic algorithm operations


Evaluation of individuals

-All
in the populationSolution suitability values of
chromosomesis calculated.

-Eachproblem specific evaluation functionmust be used.

-Problem-oriented workingthe most important partis the


evaluation function.

-Evaluationin every new generationIt is made for all


individuals.

12
Genetic algorithm operations
Creating a new population
-selecting individuals

-Selected individuals
pairing
-Two selected
individuals cross
-elitism

-in any gene of an


individualmutation
random by process
making changes

13

Genetic algorithm operations


Individual selection – deterministic

-A certain number of people in the populationA new population is


created with the best individuals.

-a certain population in the populationbelow thresholdhave value individuals


are not transferred to the new population.

-in the populationworsta certain percentagethe individual is not


transferred to the new population.

14
Genetic algorithm operations
Selecting individuals - roulette wheel
-Eachindividualto the solutionas the degree of suitability increasesto the new
populationThe chance of transmission increases.

-Individualson the roulette wheelThey have the chance to be selected


according to the area they cover.

15

Genetic algorithm operations


Selecting individuals - roulette wheel
-Eachindividualto the solutionas the degree of suitability increasesto the new
populationThe chance of transmission increases.

Beginning

B1=30% Movement

Individual 1: f(x) = 45 B2=14%


Individual 2: f(x) = 21 B4=50%
Individual 3: f(x) = 9
Individual 4: f(x) = 75 B3=6%

Individual 1: 45 / (21+75+9+45) = 30%


Individual 2: 21 / (21+75+9+45) = 14%
Individual 3: 9 / (21+75+9+45) = 6 %
Individual 4: 75 / (21+75+9+45) = 50%

16
Genetic algorithm operations
Selecting individuals - randomly

-The individual's solutionThe degree of eligibility does not affect their chances of
being selected.

Beginning

Individual 1: f(x) = 45%


Individual 2: f(x) = 21%
B1=25% Movement
Individual 3: f(x) = 9%
Individual 4: f(x) = 75%
B2=25% B4=25%

B3=25%

17

Genetic algorithm operations


Individual selection - tournament

-randomly selectedthe one with the higher fitness degree out of two
individualsis passed on to the next population.

-The process is repeated as many chromosomes as there are in the


population.

18
Contents

-genetic algorithm
-Genetic algorithm operations
-Creating the new generation
-algorithm

-Performance

19

Creating the new generation


elitism

-A certain number of the best individualswithout any processing is


transferred directly to the new generation.

-As the number of elite individuals increasesin solutiondiversity decreases.

-If an elite individual is not recruitedin the new generationbest individualOne


He may be worse than the best individual of the previous generation.

20
Creating the new generation
Cross

-The crossover points are chosen randomly.

-newlyacquiredBoth individuals can be transferred to the new


population.

-Fromthe newly acquired individualIt can be transferred to a new


population with a higher fitness level.

21

Creating the new generation


Crossover - continued
-One point crossover

0111101 0111011
Two new individuals

0111011 0111101

-Multipoint crossover

0111101 0111001
Two new individuals

0111011 0111111

22
Creating the new generation
Mutation
-in
any chromosomeIt is a completely random change of a
gene.
-The mutation rate is usually chosen between 0.01 and
0.0001.
-inthe new populationfor each gene of an individualat the given rate
There is a possibility of mutation.
-MutationIt adds diversity to the population.

0111101 0111001

23

Contents

-genetic algorithm
-Genetic algorithm operations
-Creating the new generation
-algorithm

-Performance

24
algorithm
Working of the algorithm
one.Create the initial population.

2.Evaluate each chromosome in the population with its fitness


function.

3.Create new chromosomes for the new population.


(Selection, Crossover, Mutation, Elitism operators are used.)

4.Repeat steps 2 and 3 until a suitable solution is found


or the specified iteration is reached.

5.Take the best chromosome as the solution.

25

algorithm

26
Contents

-genetic algorithm
-Genetic algorithm operations
-Creating the new generation
-algorithm

-Performance

27

Performance
Factors affecting performance
-chromosome number
-Increasing the number of chromosomes increases the working time.
-Reducing it destroys chromosome diversity.
-Mutation Rate
-When the chromosomes start to resemble each other, if a suitable
solution is still not found, it may be stuck in a certain region.
-If the best solution has not been improved for a while, it may be
stuck in a local minimum.
-mutation processIt is the only way for the algorithm to get out of
a stuck place.
-HoweverhighOnevalueto giveIt prevents the genetic
algorithm from reaching a stable point.
-mutation rate1/L(L = Number of genes in the chromosome) can be taken.
28
Performance
Factors affecting performance
-How many points will be crossed?
-NormallyThe crossover is performed at a single point.
-Studies have revealed some problemsshowed that
multipoint crossover is useful.
-Howto evaluate the individuals obtained as a
result of crossover
-The two individuals obtained may have some problems.The good ones are
transferred to the new population.

-In some problemstransferring both to the new


populationgave more successful results.
-crossover rate%95-99%receivable.
-In multi-point crossovercrossover points can be taken as
variable.
29

Performance
Factors affecting performance
-Conductingconformity assessment
-problem specificand trueneeds to be created.
-Not properly createdevaluation functionmay extend the
working time.
-Not properly createdevaluation functionIt may result in a
solution never being reached.

30
Homework

-Prepare an article research paper that includes the application of the genetic
algorithm in the fields of software engineering (software testing, etc.),
computer networks (routing, etc.) or information security (cryptology, etc.).

31

You might also like