10
Evolutionary Computation Stephanie Forrest External Professor, Santa Fe Institute Computer Science Professor, Arizona State University

evolutionary computation ool...Example Performance Curve 0 10 20 30 40 50 60 70 80 90 100 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460

  • Upload
    others

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

Page 1: evolutionary computation ool...Example Performance Curve 0 10 20 30 40 50 60 70 80 90 100 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460

Evolutionary Computation

Stephanie ForrestExternal Professor, Santa Fe InstituteComputer Science Professor, Arizona State University

Page 2: evolutionary computation ool...Example Performance Curve 0 10 20 30 40 50 60 70 80 90 100 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460

Basic Principles of Evolution • Random variation • Selection• Inheritance

Page 3: evolutionary computation ool...Example Performance Curve 0 10 20 30 40 50 60 70 80 90 100 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460

What are the genetic representations?

• Discrete units (genes) • Genotype vs. phenotype• Information stored in a linear array

How is it implemented?

Discrete units (genes)Genotype vs. PhenotypeInformation stored in a linear array

Mendel (1865)

Mendel (1865)

Crick & Watson (1953)

13

Mendel:! - Observed that traits (properties) of peas were discrete, that they came in pairs, and that one trait was often dominant.

After 85 years, the mechanism of genetics was revealed by Crick and Watson in 1953.The double helix, which for our purpose will be a linear array of genes.

Mendel (1865) Crick & Watson (1953)

Page 4: evolutionary computation ool...Example Performance Curve 0 10 20 30 40 50 60 70 80 90 100 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460

Darwinian evolution in a computer

• Random variation • Selection• Inheritance

Holland (1962)

Crossover

001111110001010

...

111001110001010

...

001111101001100

...

Selection

MutationPopulation at TN Population at TN+1

F(00111) = 0.1F(11100) = 0.9F(01010) = 0.5

Genetic Algorithm

Page 5: evolutionary computation ool...Example Performance Curve 0 10 20 30 40 50 60 70 80 90 100 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460

Example Performance Curve

0

10

20

30

40

50

60

70

80

90

100

0 20 40 60 80 100120140160180200220240260280300320340360380400420440460480500520535

Generation

Fitness

mean fitness max fitness

Page 6: evolutionary computation ool...Example Performance Curve 0 10 20 30 40 50 60 70 80 90 100 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460

Applications• Engineering

• Multi-parameter function optimization

Dimension 1

Dimen

sion 2

Charles R. Marshall (2014)

Page 7: evolutionary computation ool...Example Performance Curve 0 10 20 30 40 50 60 70 80 90 100 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460

Function Optimization•

0 0 1 1 0 1 Base 2

1 5 Base 10

F(x, y) = yx2 - x4

F(001111) = F(1,5) = 5 • 12-14 = 4

Page 8: evolutionary computation ool...Example Performance Curve 0 10 20 30 40 50 60 70 80 90 100 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460

Origins • Genetic Algorithms (Holland, 1962)• Evolutionstrategie (Rechenberg, 1965)• Evolutionary Programming (Fogel et al., 1966)• Genetic Programming (Koza, 1992)•• Evolutionary Computation

Page 9: evolutionary computation ool...Example Performance Curve 0 10 20 30 40 50 60 70 80 90 100 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460

Further Reading• An Introduction to Genetic Algorithms

Melanie Mitchell, MIT Press 1996

Introduction to Evolutionary ComputingA.E. Eiben and J.E. Smith, Springer 2015

Adaptation in Natural and Artificial Systems: An Introductory Analysiswith Applications to Biology, Control and Artificial IntelligenceJ. H. Holland, MIT Press 1992

S. Forrest. Genetic algorithms: principles of natural selection appliedto computation. Science, 261:872–878, Aug. 1993.

Page 10: evolutionary computation ool...Example Performance Curve 0 10 20 30 40 50 60 70 80 90 100 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460

ReferencesHardesty, L. Automatic Bug Repair. MIT News. 2015. Ma S, Kemmeren P, Gresham D, Statnikov A (2014) De-Novo Learning of

Genome-Scale Regulatory Networks in S. cerevisiae. PLoS ONE 9(9): e106479. https://doi.org/10.1371/journal.pone.0106479

Marshall Charles R. (2014) The evolution of morphogenetic fitness landscapes: conceptualising the interplay between the developmental and ecological drivers of morphological innovation. Australian Journal of Zoology 62, 3-17.