genetic algorithm selection

Upload: indra-mukherjee

Post on 05-Apr-2018

257 views

Category:

Documents


1 download

TRANSCRIPT

  • 7/31/2019 Genetic Algorithm Selection

    1/28

    Genetic Algorithm SelectionSchemes

    By: Mohammed MazaidaSupervisor : Dr. Eyas Alqawasmah

  • 7/31/2019 Genetic Algorithm Selection

    2/28

    Genetic Algorithm SelctionSchemes2

    1. Introduction

    Genetic Algorithms (GA) are probabilistic searchalgorithms, based on the model of natural evolution.

    GA follow the idea of SURVIVAL OF THE FITTEST-Better and better solutions evolve from previousgenerations until a near optimal solution is obtained .

    In computer world, genetic material (chromosomes)is replaced by strings of bits and natural selectionreplaced by fitness function.

  • 7/31/2019 Genetic Algorithm Selection

    3/28

    Genetic Algorithm SelctionSchemes3

    2. Basic Genetic Algorithm

    1. Starting with an initial random n population whichare suitable solutions for the problem.

    2. Evaluate the fitness of each individual ( solution ) inthe population. The fitness is a metric to indicatehow good an individual represents a solution of the

    problem3. New individuals are introduced by crossover

    operation, where at least two individuals of ageneration are chosen as 'parents'. Their genomes

    are combined to produce children.

  • 7/31/2019 Genetic Algorithm Selection

    4/28

    Genetic Algorithm SelctionSchemes4

    Basic Genetic Algorithm ,,,cont

    4. Mutation operation is performed by replacing apart of the child with a random value.

    5. Select two parent chromosomes from a populationaccording to their fitness (the better fitness, thebigger chance to be selected) The idea is to

    choose the better parents.6. Loop from step 2 until the end condition is satisfied,

    and return the best solution in current population,

  • 7/31/2019 Genetic Algorithm Selection

    5/28

    Basic Genetic Algorithm ,,,cont

    Describe

    Problem

    Generate

    Initial

    Solutions

    Test: is initial

    solution good enough?Stop

    Select parents

    to reproduce

    Apply crossover process

    and create a set of offspring

    Apply random mutation

    Yes

    No

  • 7/31/2019 Genetic Algorithm Selection

    6/28

  • 7/31/2019 Genetic Algorithm Selection

    7/28

    Genetic Algorithm SelctionSchemes7

    Selection schemescont

    Selection schemes characterized by several terms such as :

    selection intensity IP expected average fitness value ofthe population after applying a selection method .

    Selection variance is the expected variance of the fitnessdistribution of the population after applying the selectionmethod to the normalized Gaussian distribution

    Loss of diversity proportion of individuals of a populationthat is not selected during the selection phase[1]

  • 7/31/2019 Genetic Algorithm Selection

    8/28

    Genetic Algorithm SelctionSchemes8

    The Selection schemes ,,,cont

    Commonly used selection schemes:

    Proportionate reproduction. Tournament selection.

    Truncation selection

    Linear ranking selection.

  • 7/31/2019 Genetic Algorithm Selection

    9/28

    Genetic Algorithm SelctionSchemes9

    3.1 Proportional Selection

    also known as roulette-wheel selection

    Parents are selected according to their fitness. Thebetter , the more chances to be selected

    Can be imagined as roulette wheel where areplaced all individuals, every has its place bigaccordingly to its fitness function.

    Then a marble is thrown there and selects theindividual . individuals with bigger fitness will beselected more times.

  • 7/31/2019 Genetic Algorithm Selection

    10/28

    Genetic Algorithm SelctionSchemes10

    Proportional Selectioncont

    This can be simulated by following algorithm.

    [Sum] Calculate sum of all chromosome fitnesses in

    population - sum S. [Select] Generate random number from interval

    (0,S)- r.

    [Loop] Go through the population and sum fitnesses

    from 0 - sum s. When the sum s is greater then r,stop and return the chromosome where you are.

    Of course, step 1 is performed only once for eachpopulation.

  • 7/31/2019 Genetic Algorithm Selection

    11/28

    Genetic Algorithm SelctionSchemes11

    Proportional Selectioncont

    individual i in the population, its probability of beingselected is

    where N is the number of individuals in thepopulation.

    The time complexity of the algorithm is O(N).

    Selection intensity

    where is the mean variance of the fitness values ofthe population before selection.

  • 7/31/2019 Genetic Algorithm Selection

    12/28

    Genetic Algorithm SelctionSchemes12

    Proportional Selectioncont

    roulette-wheel selection algorithm [2]

  • 7/31/2019 Genetic Algorithm Selection

    13/28

    Genetic Algorithm SelctionSchemes13

    3.2 Tournament selection

    Choose some number of individuals ( t) (tournamentsize )randomly from the population and select the best

    individual from this group as parent.

    Repeat N time.

    No sorting of population is needed

    It has the time complexity O(N).

  • 7/31/2019 Genetic Algorithm Selection

    14/28

    Genetic Algorithm SelctionSchemes14

    Tournament selectioncont

    Selection intensity

    Selection variance:

    This method is not very useful when we use largepopulation because we will need a lot of N time tosearch every time on new element from the selectedgroup randomly

  • 7/31/2019 Genetic Algorithm Selection

    15/28

  • 7/31/2019 Genetic Algorithm Selection

    16/28

    Genetic Algorithm SelctionSchemes16

    Tournament selection ...cont

    Gaussian fitness distribution approximately leads again toGaussian distributions after tournament selection (from leftto right: initial distribution, t =2, t = 5, t = 10).[2]

  • 7/31/2019 Genetic Algorithm Selection

    17/28

    Genetic Algorithm SelctionSchemes17

    Tournament selection ...cont

    Properties of tournament selection[1]

  • 7/31/2019 Genetic Algorithm Selection

    18/28

    Genetic Algorithm SelctionSchemes18

    3.3 Truncation selection

    Individuals are sorted according to their fitness.

    Only the individuals above the threshold T areselected as parents.

    As a sorting of the population is required, truncationselection has a time complexity of O(N ln N).

  • 7/31/2019 Genetic Algorithm Selection

    19/28

    Genetic Algorithm SelctionSchemes19

    Truncation selectioncont

    Selection intensity

    Selection Variance:

  • 7/31/2019 Genetic Algorithm Selection

    20/28

    Genetic Algorithm SelctionSchemes20

    Truncation selectioncont

    Truncation selection algorithm [2]

  • 7/31/2019 Genetic Algorithm Selection

    21/28

    Genetic Algorithm SelctionSchemes21

    Truncation selectioncont

    Properties of truncation selection[1]

  • 7/31/2019 Genetic Algorithm Selection

    22/28

    Genetic Algorithm SelctionSchemes22

    3.4 Linear Ranking Selection

    For ranking selection the individuals are sortedaccording their fitness values and the rank N is

    assigned to the best individual and the rank 1 to theworst individual.

    The probability for the individual to be selected givenby

  • 7/31/2019 Genetic Algorithm Selection

    23/28

    Genetic Algorithm SelctionSchemes23

    Linear Ranking Selectioncont

    As a sorting of the population is required, it

    has a time complexity of O(N log N). Selection intensity

    Where /N is the probability of the worstindividual to be Selected

  • 7/31/2019 Genetic Algorithm Selection

    24/28

    Genetic Algorithm SelctionSchemes24

    Linear Ranking Selectioncont

    Linear ranking selection algorithm[2]

  • 7/31/2019 Genetic Algorithm Selection

    25/28

    Genetic Algorithm SelctionSchemes25

    Linear Ranking Selectioncont

    Properties of linear ranking[1]

  • 7/31/2019 Genetic Algorithm Selection

    26/28

    Genetic Algorithm SelctionSchemes26

    Linear Ranking Selectioncont

    Gaussian fitness distribution and the resulting distributionsafter performing ranking selection with - = 0.5 and - = 0(from left to right).[2]

  • 7/31/2019 Genetic Algorithm Selection

    27/28

    Genetic Algorithm SelctionSchemes27

    Conclusion

    A genetic algorithm conceptually a simulation ofsteps inspired by the biological processes of

    evolution. Selection schemes are use to choose the

    individuals in the population that will create offspringfor the next generation.

    Proportionate reproduction, Tournament selection,

    Truncation selection and Linear ranking selectionare examples of selection schemes Selection schemes are characterized by selection

    intensity, selection variance and loss of diversity.

  • 7/31/2019 Genetic Algorithm Selection

    28/28

    Genetic Algorithm SelctionSchemes28

    Reference

    1. http://www.geatbx.com/docu/algindex-02.html

    2. Blickle,T.,& Thiele,L.(1995). A comparison of selection

    schemes used in genetic algorithms. (TechnicalReportNo.11).Gloriastrasse35,CH-8092 Zurich: Swiss Federal Institute ofTechnology (ETH) Zurich, Computer Engineering andCommunications Networks Lab(TIK).

    3. Stuart Russell & Peter Norvig. Artificial Intelligence: A ModernApproach. 2 Edition / 2003

    4. http://www.talkorigins.org/faqs/genalg/genalg.html5. BradL.Miller & DavidE.Goldberg (1996). Genetic Algorithms,

    Selection Schemes, and the Varying effects of Noise.

    http://www.geatbx.com/docu/algindex-02.htmlhttp://www.talkorigins.org/faqs/genalg/genalg.htmlhttp://www.talkorigins.org/faqs/genalg/genalg.htmlhttp://www.talkorigins.org/faqs/genalg/genalg.htmlhttp://www.talkorigins.org/faqs/genalg/genalg.htmlhttp://www.talkorigins.org/faqs/genalg/genalg.htmlhttp://www.talkorigins.org/faqs/genalg/genalg.htmlhttp://www.talkorigins.org/faqs/genalg/genalg.htmlhttp://www.talkorigins.org/faqs/genalg/genalg.htmlhttp://www.talkorigins.org/faqs/genalg/genalg.htmlhttp://www.talkorigins.org/faqs/genalg/genalg.htmlhttp://www.geatbx.com/docu/algindex-02.htmlhttp://www.geatbx.com/docu/algindex-02.htmlhttp://www.geatbx.com/docu/algindex-02.htmlhttp://www.geatbx.com/docu/algindex-02.htmlhttp://www.geatbx.com/docu/algindex-02.htmlhttp://www.geatbx.com/docu/algindex-02.htmlhttp://www.geatbx.com/docu/algindex-02.htmlhttp://www.geatbx.com/docu/algindex-02.htmlhttp://www.geatbx.com/docu/algindex-02.htmlhttp://www.geatbx.com/docu/algindex-02.htmlhttp://www.geatbx.com/docu/algindex-02.html