soft computing evolutionary computing

45
1 SOFT COMPUTING Evolutionary Computing

Upload: june

Post on 22-Feb-2016

73 views

Category:

Documents


2 download

DESCRIPTION

SOFT COMPUTING Evolutionary Computing. EC. What is a GA?. GAs are adaptive heuristic search algorithm based on the evolutionary ideas of natural selection and genetics. As such they represent an intelligent exploitation of a random search used to solve optimization problems. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: SOFT COMPUTING Evolutionary Computing

11

SOFT COMPUTINGEvolutionary Computing

Page 2: SOFT COMPUTING Evolutionary Computing

What is a GA? GAs are adaptive heuristic search algorithm based on the

evolutionary ideas of natural selection and genetics.

As such they represent an intelligent exploitation of a random search used to solve optimization problems.

Although randomized, GAs are by no means random, instead they exploit historical information to direct the search into the region of better performance within the search space.

Page 3: SOFT COMPUTING Evolutionary Computing

What is a GA? The basic techniques of the GAs are designed to simulate

processes in natural systems necessary for evolution, specially those follow the principles first laid down by Charles Darwin of "survival of the fittest.".

Since in nature, competition among individuals for scanty resources results in the fittest individuals dominating over the weaker ones.

Page 4: SOFT COMPUTING Evolutionary Computing

Evolutionary Algorithms

Genetic Programming

Evolution Strategies

Genetic Algorithms

EvolutionaryProgramming

Classifier Systems

• genetic representation of candidate solutions• genetic operators• selection scheme• problem domain

Page 5: SOFT COMPUTING Evolutionary Computing

History of GAs Genetic Algorithms were invented to mimic some of the

processes observed in natural evolution. Many people, biologists included, are astonished that life at the level of complexity that we observe could have evolved in the relatively short time suggested by the fossil record.

The idea with GA is to use this power of evolution to solve optimization problems. The father of the original Genetic Algorithm was John Holland who invented it in the early 1970's.

Page 6: SOFT COMPUTING Evolutionary Computing

Classes of Search Techniques

F inon acc i N ew ton

D irect m ethods Indirec t m ethods

C alcu lus-based tech n iqu es

Evolu tionary s trategies

C e ntra l ized D is tr ibuted

Paral le l

S tea dy-s tate G enerationa l

S equ entia l

G ene tic a lgor ithm s

E volu tiona ry a lgori thm s S im u lated annealing

G uided rand om search techniques

D ynam ic program m ing

E num erative techn iques

S earch te chniques

Tabu Search Hill Climbing

DFS, BFS

Genetic Programming

Page 7: SOFT COMPUTING Evolutionary Computing

Early History of EAs 1954: Barricelli creates computer simulation of life – Artificial Life 1957: Box develops Evolutionary Operation (EVOP), a non-computerised

evolutionary process 1957: Fraser develops first Genetic Algorithm 1958: Friedberg creates a learning machine through evolving computer

programs 1960s, Rechenverg: evolution strategies

a method used to optimize real-valued parameters for devices 1960s, Fogel, Owens, and Walsh: evolutionary programming

to find finite-state machines 1960s, John Holland: Genetic Algorithms

to study the phenomenon of adaptation as it occurs in nature (not to solve specific problems)

1965: Rechenberg & Schwefel independently develop Evolution Strategies 1966: L. Fogel develops Evolutionary Programming as a means of creating

artificial intelligence 1967: Holland and his students extend GA ideas further

Page 8: SOFT COMPUTING Evolutionary Computing

The Genetic Algorithm Directed search algorithms based on the mechanics of

biological evolution Developed by John Holland, University of Michigan (1970’s)

To understand the adaptive processes of natural systems To design artificial systems software that retains the

robustness of natural systems

The genetic algorithms, first proposed by Holland (1975), seek to mimic some of the natural evolution and selection.

The first step of Holland’s genetic algorithm is to represent a legal solution of a problem by a string of genes known as a chromosome.

Page 9: SOFT COMPUTING Evolutionary Computing

Evolutionary Programming First developed by Lawrence Fogel in 1966 for use in

pattern learning Early experiments dealt with a number of Finite State

Automata FSA were developed that could recognise recurring

patterns and even primeness of numbers Later experiments dealt with gaming problems

(coevolution) More recently has been applied to training of neural

networks, function optimisation & path planning problems

Page 10: SOFT COMPUTING Evolutionary Computing

Biological Terminology• gene

• functional entity that codes for a specific feature e.g. eye color• set of possible alleles

• allele• value of a gene e.g. blue, green, brown• codes for a specific variation of the gene/feature

• locus• position of a gene on the chromosome

• genome• set of all genes that define a species• the genome of a specific individual is called genotype• the genome of a living organism is composed of several chromosomes

• population• set of competing genomes/individuals

Page 11: SOFT COMPUTING Evolutionary Computing

Genotype versus Phenotype• genotype

• blue print that contains the information to construct an organism e.g. human DNA• genetic operators such as mutation and recombination modify the genotype during reproduction• genotype of an individual is immutable

(no Lamarckian evolution)

• phenotype• physical make-up of an organism• selection operates on phenotypes (Darwin’s principle : “survival of the fittest”)

Page 12: SOFT COMPUTING Evolutionary Computing

Courtesy of U.S. Department of Energy Human Genome Program , http://www.ornl.gov/hgmis

Page 13: SOFT COMPUTING Evolutionary Computing

Genotype Operators• recombination (crossover)

• combines two parent genotypes into a new offspring• generates new variants by mixing existing genetic material• stochastic selection among parent genes

• mutation• random alteration of genes• maintain genetic diversity

• in genetic algorithms crossover is the major operator whereas mutation only plays a minor role

Page 14: SOFT COMPUTING Evolutionary Computing

Crossover• crossover applied to parent strings with probability pc : [0.6..1.0]• crossover site chosen randomly• one-point crossover

parent Aparent B

1 1 0 1 01 0 0 0 1

offspring Aoffspring B

1 1 0 1 1

1 0 0 0 0

• two-point crossoverparent Aparent B

1 1 0 1 01 0 0 0 1

offspring Aoffspring B

1 1 0 0

1 0 0 1

0

1

Page 15: SOFT COMPUTING Evolutionary Computing

Mutation

• mutation applied to allele/gene with probability Pm : [0.001..0.1]• role of mutation is to maintain genetic diversity

offspring: 1 1 0 0 0

Mutate fourth allele (bit flip)

1 1 0 0 01mutated offspring:

Page 16: SOFT COMPUTING Evolutionary Computing

recombination

10001

01011

10011

01001

mutation

x

f

phenotype space

Structure of an Evolutionary Algorithm

00111 11001

10001

01011

population of genotypes

coding scheme

fitness

selection

11001

10001

0101110001

10111

01001

10

01

001

011

10

01 001

01110011

01001

Page 17: SOFT COMPUTING Evolutionary Computing

Pseudo Code of an Evolutionary Alg.Create initial random population

Evaluate fitness of each individual

Termination criteria satisfied ?

Select parents according to fitness

Recombine parents to generate offspring

Mutate offspring

Replace population by new offspring

stopyes

no

Page 18: SOFT COMPUTING Evolutionary Computing

Roulette Wheel Selection

• probability of reproduction pi = fi / Sk fk

10001

11010

01011

00101

10001

11010

01011

0010110001

11010

01011

00101

10001

11010

01011

00101

10001 11010

01011

00101

• selected parents : 01011, 11010, 10001, 10001

• selection is a stochastic process

Page 19: SOFT COMPUTING Evolutionary Computing

• automatic generation of computer programs by means of natural evolution see Koza 1999• programs are represented by a parse tree (LISP expression)• tree nodes correspond to functions : - arithmetic functions {+,-,*,/} - logarithmic functions {sin,exp}• leaf nodes correspond to terminals : - input variables {X1, X2, X3} - constants {0.1, 0.2, 0.5 }

Genetic Programming

+*

X3X2

X1

tree is parsed from left to right: (+ X1 (* X2 X3)) X1+(X2*X3)

Page 20: SOFT COMPUTING Evolutionary Computing

Genetic Programming : Crossover

+*

X3X2

X1

-/

X1-X2

X3X2

parent A parent B

+ *X3X2X1

-/

X1-X2

X3X2

offspring A offspring B

Page 21: SOFT COMPUTING Evolutionary Computing

Areas EAs Have Been Used InDesign of electronic circuitsDesign of electronic circuitsTelecommunication network Telecommunication network designdesignArtificial intelligenceArtificial intelligenceStudy of atomic clustersStudy of atomic clustersStudy of neuronal behaviourStudy of neuronal behaviourNeural network training & designNeural network training & designAutomatic controlAutomatic controlArtificial lifeArtificial lifeSchedulingScheduling

Travelling Salesman ProblemTravelling Salesman ProblemGeneral function optimisation General function optimisation Bin Packing ProblemBin Packing ProblemPattern learningPattern learningGamingGamingSelf-adapting computer programsSelf-adapting computer programsClassificationClassificationTest-data generationTest-data generationMedical image analysisMedical image analysisStudy of earthquakesStudy of earthquakes

Page 22: SOFT COMPUTING Evolutionary Computing

Goldberg (1989)

Goldberg D. E. (1989), Genetic Algorithms in Search, Optimization, and Machine Learning. Addison-Wesley, Reading.

Page 23: SOFT COMPUTING Evolutionary Computing

Michalewicz (1996)

Michalewicz, Z. (1996), Genetic Algorithms + Data Structures = Evolution Programs, Springer.

Page 24: SOFT COMPUTING Evolutionary Computing

Vose (1999)

Vose M. D. (1999), The Vose M. D. (1999), The Simple Genetic Simple Genetic Algorithm : Algorithm : Foundations and Foundations and Theory (Complex Theory (Complex Adaptive Systems). Adaptive Systems). Bradford Books; Bradford Books;

Page 25: SOFT COMPUTING Evolutionary Computing

2525

SOFT COMPUTINGFuzzy-Evolutionary Computing

Page 26: SOFT COMPUTING Evolutionary Computing
Page 27: SOFT COMPUTING Evolutionary Computing
Page 28: SOFT COMPUTING Evolutionary Computing
Page 29: SOFT COMPUTING Evolutionary Computing

Genetic Fuzzy Systems (GFS’s)

• genetic design of fuzzy systems• automated tuning of the fuzzy knowledge base• automated learning of the fuzzy knowledge base• objective of tuning/learning process

• optimizing the performance of the fuzzy system: e.g.: fuzzy modeling : minimizing quadratic error between data set and the fuzzy system outputs e.g : fuzzy control system: optimize the behavior of the plant + fuzzy controller

Page 30: SOFT COMPUTING Evolutionary Computing

Genetic Fuzzy System for Data Modeling

Evolutionaryalgorithm

Evaluationscheme

Dataset : xi,yiFuzzy System

Fuzzy system parameters

fitness

genotype

phenotype

Page 31: SOFT COMPUTING Evolutionary Computing

Fuzzy SystemsKnowledge Base

Database :Definition of

fuzzy membership-function

a b c

Rule base:definition offuzzy rules

If X1 is A1 and … and Xn is An

then Y is B

Page 32: SOFT COMPUTING Evolutionary Computing

Genetic Tuning Process• tuning problems utilize an already existing rule base• tuning aims to find a set of optimal parameters for the database :

• points of membership-functions [a,b,c,d] or• scaling factors for input and output variables

Page 33: SOFT COMPUTING Evolutionary Computing

Linear Scaling FunctionsChromosome for linear scaling:• for each input xi : two parameters ai,bi i=1..n• for the output y : two parameter a0,b0

Genetic Algorithms: • encode each parameter by k bit using Gray code total length = 2*(n+1)*k bit

Evolutionary Strategies:• each parameter ai or bi corresponds to one object variable xm m : 1… 2*(n+1)

a0100101

b0011111

a1110101

b2*(n+1)100101. . .

a0x0,o

b0x1,1

a1x2,2

b2*(n+1) xm,m. . .

Page 34: SOFT COMPUTING Evolutionary Computing

Descriptive Knowledge Base• descriptive knowledge base

sm me lg

x

neg ze pos

y

• all rules share the same global membership functions :

R1 : if X is sm then Y is neg R2 : if X is me then Y is ze R3 : if X is lg then Y is pos

Page 35: SOFT COMPUTING Evolutionary Computing

Approximate Knowledge Base• each rule employs its own local membership function

R1 : if X is then Y is R1 : if X is then Y is R1 : if X is then Y is R1 : if X is then Y is

• tradeoff: more degrees of freedom and therefore better approximation but intuitive meaning of fuzzy sets gets lost

Page 36: SOFT COMPUTING Evolutionary Computing

Tuning Membership Functions• encode each fuzzy set by characteristic parameters

x

(x)1

0 a b c d

Trapezoid: <a,b,c,d>

x

(x)1

0

Gaussian: N(m,s)

m

s

xx

(x)1

0 a b c

Triangular: <a,b,c>

Page 37: SOFT COMPUTING Evolutionary Computing

Approximate Genetic Tuning Process• a chromosome encodes the entire knowledge base, database and rulebaseRi : if x1 is Ai1 and … xn is Ain then y is Bi

encoded by the i-th segment Ci of the chromosomeusing triangular membership-functions (a,b,c)

(ai1, bi1, ci1, . . . , ain, bin, cin, ai, bi, ci, )Ci =

The chromosome is the concatenation of the individual segments corresponding to rules :C1 C2 C3 C4 . . . Ck

each parameter may be binary or real-coded

Page 38: SOFT COMPUTING Evolutionary Computing

Descriptive Genetic Tuning Process• the rule base already exists• assume the i-th variable is composed of N i terms

(ai1, bi1, ci1, . . . , aiNi, biNi, ciNi )Ci =

A1 A2 A3

xiai1, bi1, ci1, ai2, bi2, ci2 ai3, bi3, ci3 The chromosome is the concatenation of the individual segments corresponding to variables :

C1 C2 C3 C4 . . . Ck

Page 39: SOFT COMPUTING Evolutionary Computing

Descriptive Genetic Tuning• in the previous coding scheme fuzzy sets might change their order and optimization is subject to the constraints : aij < bij < cij

A1 A2 A3

x1 x2 x3

• encode the distance among the center points of triangular fuzzy sets and choose the border points such that i = 1

Page 40: SOFT COMPUTING Evolutionary Computing

Fitness Function for Tuning

• minimize quadratic error among training data (xi,yi) and fuzzy system output f(xi) E = Sumi (yi-f(xi))2

Fitness = 1 / E (maximize fitness)

• minimize maximal error among training data (xi,yi) and fuzzy system output f(xi) E = maxi (yi-f(xi))2

Fitness = 1 / E (maximize fitness)

Page 41: SOFT COMPUTING Evolutionary Computing

Genetic Learning Systems• genetic learning aim to :

• learn the fuzzy rule base or• learn the entire knowledge base

• three different approaches• Michigan approach : each chromosome represents a single rule• Pittsburgh approach : each chromosome represents an entire rule base / knowledge base• Iterative rule learning : each chromosome represents a single rule, but rules are injected one after the other into the knowledge base

Page 42: SOFT COMPUTING Evolutionary Computing

Michigan Approach

11001 : R1: if x is A1 ….then Y is B100101 : R2: if x is A2 ….then Y is B210111 : R3: if x is A3 ….then Y is B311100 : R4: if x is A4 ….then Y is B401000 : R5: if x is A5 ….then Y is B511101 : R6: if x is A6 ….then Y is B6

Population:Individual:

A1A3A4

A2A5A6

XB1

B3

B4

B2

B5

B6

Y

Page 43: SOFT COMPUTING Evolutionary Computing

R4 : if x is small then Y is pos.F=-1.6

R1 : if x is large then Y is neg.F = 2.5

R3 : if x is small then Y is zeroF=-0.4

R2 : if x is med. then Y is zeroF=2.7

Cooperation vs. Competition Problem• we need a fitness function that measures the accuracy of an individual rule as well as the quality of its cooperation with other rules

X

Y

small medium large

pos

zene

g

Fitness = number of correct classifications minus number of incorrect classifications

Page 44: SOFT COMPUTING Evolutionary Computing

Michigan Approach• steady state selection:

• pick one individual at random• compare it with all individuals that cover the same input region• remove the “relatively” worst one from the population • pick two parents at random independent of their fitness and generate a new offspring11001 : R1: if x is A1 ….then Y is B1

00101 : R2: if x is A2 ….then Y is B2

10111 : R3: if x is A3 ….then Y is B3

11100 : R4: if x is A4 ….then Y is B4

01000 : R5: if x is A5 ….then Y is B5

11101 : R6: if x is A6 ….then Y is B6

11001 : R1: if x is A1 ….then Y is B110111 : R3: if x is A3 ….then Y is B3

11100 : R4: if x is A4 ….then Y is B4

competitors:

removed from the population

Page 45: SOFT COMPUTING Evolutionary Computing

4545

Thanks for your attention!Thanks for your attention!

That’s all.That’s all.