genetic algorithm ( ga )

18

Click here to load reader

Upload: abuamo

Post on 20-Jun-2015

402 views

Category:

Technology


0 download

TRANSCRIPT

Page 1: GENETIC ALGORITHM ( GA )

GENETIC ALGORITHM ( GA )

Page 2: GENETIC ALGORITHM ( GA )

Introduction

As we increase the number of objectives we are trying to achieve we also increase the number of constrains on the problem and complixity increases ..

Genatic algorithm GA is ideal for these types of problems where the search space is large & number of feasiables solutions is small

Page 3: GENETIC ALGORITHM ( GA )

Introduction

It’s an adaptive heurastic search algorithm based on the evoultionary ideas of natural selections and genatic

It follows the principles laid down by darwin of survival to the fittest

A fittest fuction is used to evaluate indviduales

Page 4: GENETIC ALGORITHM ( GA )

Technique of GA

• generate intial population M(0)

• compute and save the fitness U(M) of each indvidual m in population M(T)

• define selection probablties P(M) for each indvidual M in M(T) so that P(M) is propotional to U(M)

• generate M(T+1) by selecting indviduals from M(T) to mate

• step 2 repeated untill desired character ( sutiable solution ) obtained

Page 5: GENETIC ALGORITHM ( GA )

Methodology

Intialization

Selection

Genetic operator

Termination

Page 6: GENETIC ALGORITHM ( GA )

Methodology

• solution are represented in binary as a string of 0s & 1s

• the evolution usually starts from a population of randomly generated

Initialization

Selection

• in each generation the fitness of each indvidual is evaluated & inviduals selected accordingly to their fitness

Page 7: GENETIC ALGORITHM ( GA )

Methodology

• after selection an operator needs to get the second generation

•New generations should have more fitness

Genetic operator

termination

• through wich generation will stop due to : 1) fixed number of generation reached 2) maximum number of solution

Page 8: GENETIC ALGORITHM ( GA )

GA applications

Any one who can encode solutions of a given problem to chromosomes in GA & compare the performance ( fitness ) os solutions

Computer architecture : using GA to find out weak links

Learning robots behavior using genatic algorithm

Automated design of mechatronics systems using bond graphs

Page 9: GENETIC ALGORITHM ( GA )

Adaptive , fuzzy logic & heurstic GA Despite the succeful applications of

GA to numerous optimization of several problems the identification of the correct settings of genatic parameters such as ( population size , crossover & mutation operator ) is not an easy task

Many works have been performed in order to identify the correct settings value by trial & error

Page 10: GENETIC ALGORITHM ( GA )

Fuzzy logic control GA

• fuzzy is capable to adjust the rate of crossover & mutation operators

•Recently CHEONG & LAI (2000) said that GA controlled by fuzzy logic control are more efficent in search speed & search quality of GA without FLC

Page 11: GENETIC ALGORITHM ( GA )

Heurastic GA

Based on the fact that it encourages well-performed operators to produce more efficient offspring while reducing the chance of poorly performing operators to destroy the potential indviduals during genatic search process

Page 12: GENETIC ALGORITHM ( GA )

Canonical GA ( CGA )

We uses a real number representation instead of abit-string

• STEP 1 ) : intial population i.e : any random number of population

•STEP 2) : genetic operator i.e : selection , cross over & mutation

•STEP 3) : stop condition i.e : maximum number of muattion

Page 13: GENETIC ALGORITHM ( GA )

Adaptive genetic algorithm ( AGA)

STEP 1) same as CGA.

STEP 2) same as CGA .

STEP 3) Regulating GA parameters

STEP 4) repeat step 2&3 untill stop condition is reached .

Page 14: GENETIC ALGORITHM ( GA )
Page 15: GENETIC ALGORITHM ( GA )

Some observation about GA the right selection is one of the most

imoprtant genetic operators .

Learning how to tune the parameters as ( mutation , probablity , population size ...etc) is important

Because a very small mutation rate may lead to genetic drift or the high rate of mutation may lead to loss of good solutions .

Page 16: GENETIC ALGORITHM ( GA )

Simple generational genetic algorithm procedures

1) Choose intial population indvidual

2) Evaluate the fitness of that indvidual

3) Select the best fitness indvidual for matting

4) Breed new indviduals by crossing over

5) Evaluate the new indvidual fitness

Repeat till termination

Page 17: GENETIC ALGORITHM ( GA )

Criticism

• GA didn’t do well with complexity as the large number of elements exposed to mutation ie : desiging an engine .

• Sometimes the stop conditions is not clear as the better solution is only in comparison to other solution . .• Operating a dynamic data is difficult as genom start to convert and become no longer a valid data later

Page 18: GENETIC ALGORITHM ( GA )

THANK YOU

ZIAD ZOHDY