Introduction to Evolutionary Computation and Evolutionary Computation Module Tutorial 5

V. Landassuri-Moreno [e-mail]

Description

Exercise 6 EA with Real Implementation (Matlab code) SummaryIntroduction to Evolutionary Computation and EvolutionaryComputation ModuleTutorial 5V. [e-mail] of Computer ScienceUniversity of BirminghamNovember 23, 20091/11Exercise 6 EA with Real Implementation (Matlab code) SummaryOutlineExercise 6EA with Real Implementation (Matlab code)Summary2/11I Weakness:I The main issue here is that for a particular problem, you need to design a repairalgorithmI There is not a general equation that could be appliedI Depending of the problem at hand, the Repair algorithm could be as difficult as tosolve the original problemExercise 6 EA with Real Implementation (Matlab code) SummaryQ1. Weakness in a Repair AlgorithmRepair Algorithm:I Repair algorithms let us map infeasible individuals into feasible onesExercise 6 3/11Exercise 6 EA with Real Implementation (Matlab code) SummaryQ1. Weakness in a Repair AlgorithmRepair Algorithm:I Repair algorithms let us map infeasible individuals into feasible onesI Weakness:I The main issue here is that for a particular problem, you need to design a repairalgorithmI There is not a general equation that could be appliedI Depending of the problem at hand, the Repair algorithm could be as difficult as tosolve the original problemExercise 6 3/11Speciation:I Makes a parallel search in each optima to find multiple solutions at the same time(identifying the actual peak in ...

Informations

• Publié par: Iddu
• Nombre de lectures: 17
• Langue: English

Extrait

Exercise6EAwitheRlamIlpmeneatit(MonlaatodbcSue)rammy

School of Computer Science University of Birmingham

November 23, 2009

Introduction to Evolutionary Computation and Evolutionary Computation Module Tutorial 5

V. Landassuri-Moreno [e-mail]

1/11

exEmIlpmeneatitnoM(rcise6EAwithRealyodbclaatarmmSue)

code)

(Matlab

Outline

Summary

Exercise 6

Implementation

EA with Real

11/2

rcise6EAExeodbclaat(MontitanemelpmIlaeRhtiwymmra)euS maheisinehsueierahtsroftrapaucit W I aenkse:s T I erehT I mhtirogla r quleranegetanois uoenmey,orlbalprepaignardesiedto elbohtamtfogrpehaieplgrad,aneRth ctuodlebtaoitnah Dependin applied I reic me xE1/1es36 rigitheoroblnalp atlucfifi evlos otscohmitorsdeadbul

Repair Algorithm:

Repair algorithms let us map infeasible individuals into feasible ones

Weakness in a Repair Algorithm

Q1.

I

ciseExer

I Weakness: I particular problem, you need to design a repairThe main issue here is that for a algorithm I There is not a general equation that could be applied I problem at hand, the Repair algorithm could be as difficult as toDepending of the solve the original problem

Q1. Weakness in a Repair Algorithm

11/36

Repair Algorithm: I let us map infeasible individuals into feasible onesRepair algorithms

lpmeneatitnoM(taise6EAwithRealImEcrexyrammuS)edocbal

I

charea) E peakinea /411

Q2. Main differences between Niching and Speciation

Finds and maintain subpopulation of solutions around multiple optima (finding different peaks)

Niching:

excrsi6e enid e ( imetamesth lautcaehtgniyfit dmultofintimachopsntatuoiseloitlp yramum)SdecoabtlMan(emtntaoiaeIlpmel6EAwithRExercise ecSp elserallineaarch no I :aiit asapaMek

icreE6esxEammuS)edyrRhaewAtielemIlpmion(ntatabcoMatl

Speciation: I find multiple solutions at the same timeMakes a parallel search in each optima to (identifying the actual peak in each area)

Niching: I Finds and maintain subpopulation of solutions around multiple optima (finding different peaks)

11

Q2. Main differences between Niching and Speciation

Eex6e/4crsi

Landsca ogol If y I uratbialwygonhti I :ep lanA ouwiionyunctivefejtcehboolttoyputincfuhefteoapscdnaLehtsetaercll abcode)Summary I )sy oyfI v(saella ivnduaidakutnieaitlaosul(lpatoneorrespontion)itc h ehtnitniopaotsd ysrgne e E anpleryp fiifcpSce ec I :ruaf pe I C dscaflanormo toon tiopzemi he I Tuoynevahsehtcraehspaceplottedinaigevrnnaegw(ihhc im-d a n is alonsienalprepyh actI I )en esomnhavks,tepeasha,orguaterdnafl

variables

in

Fitness landscape and energy an EA and the solutions in the

the

relationships

between

1

Q3.LandscapesandEnergy surfaces

surface describe population.

65/1icesxEre itnoimaz oitulosaimretednheeter I HofgyerEn cilahpsyelsmrpboromcomefcal/hemi rperelbofosminim n I io uaUsytllyahe ehboejtcvifenutcnatethefitnessoft lementation(MatlesE6wAtiRhaeIlpmExcier

ysurnerg: I Sp faceE om I C peca hemcroef rofcfiicesdnalfom icre56es11/

Landscape: I Analogy with natural biology I If you plot the objective function you will creates the Landscape of the function to optimize I Then you have the search space plotted in a given range (which is a n-dimensional hyperplane) I It can have some peaks, troughs, and flat areas (valleys) I individual (a potential solution) it corresponds to a point in theIf you take an hyperplane

Fitness landscape and energy surface describe the relationships between variables in an EA and the solutions in the population.

Q3.LandscapesandEnergy surfaces

ivctjeobheftsoes llausU I noitcnufe lemsprobyareytheno xE azitinimfoim icyspralcamiphl/ terenEeh elbo H I sm olutiondergyofasethtfienttereiman decoum)SMan(abtltnemoitaIlaeelpm6EAwithRExerciseryma

erExE6wAicesaeIltiRhmentmplen(Maatio Exerci

Landscape: I Analogy with natural biology I If you plot the objective function you will creates the Landscape of the function to optimize I Then you have the search space plotted in a given range (which is a n-dimensional hyperplane) I It can have some peaks, troughs, and flat areas (valleys) I If you take an individual (a potential solution) it corresponds to a point in the hyperplane

Q3.LandscapesandEnergy surfaces

Energy surface: I Specific form of landscape I Come from chemical/physical problems I the Energy of a solution determinate the fitness of the objective functionHere I Usually they are problems of minimization

Fitness landscape and energy surface describe the relationships between variables in an EA and the solutions in the population.

es561/1 mary)Sumcodetlab

abcode)Summaryelemtntaoi(naMlt6EseitAweahRmplIxEicre

Q4. List of fitness sharing and implicit fitness sharing from paper:Every niching method has its niche(lecture 06)

Fitness sharing I Small population I It can find the optima with larger basins of attractions I Less distracted by peaks with small basin of attraction

icrexE ?t 11/66es mpliIsentfiticgnirahss soerov I Coramimptcemorpheneisevylevenwithsmallbasfonirttaitca I I no sntideeeladaeprgormsntofatioopulepkaaehcsetaepic wht,Bu streteinof popehtfineppahtasnotsolaulationiocevarllgroenrto ow ? Hulco epthksea nimrietauoydeted

Implicit fitness sharing I Covers optima more comprehensively even with small basin of attraction I It is needed a large population to form species at each peak of interest

Fitness sharing I Small population I It can find the optima with larger basins of attractions I Less distracted by peaks with small basin of attraction

Q4. List of fitness sharing and implicit fitness sharing from paper:Every niching method has its niche(lecture 06)

1cise66/1 eit? Exer etmrnitaluyduoed ? H ks coow thlleaepoctoarevgralnroepphaatwhepthifenoitalupoostonsinuB,t ExerlempntmeioatMan(esicwAE6RhtiIlaeltbaocedS)muamyr