johngerolecture

johngerolecture - How do genetic algorithms relate to their...

Info iconThis preview shows pages 1–6. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: How do genetic algorithms relate to their biological origins? How do they relate to human processes of design? Why is there power in this metaphor? How can they be used to extend the capabilities of the designer? John Gero Professor of Design Science University of Sydney Visiting Professor of Design and Computation MIT MIT Design Inquiry How do genetic algorithms relate to their biological origins? l l l l l Separation of genetic material (genotype [representation]) from organism (phenotype [design]) Expressing genotype as organism Organism carries genotype and reproduces genotype using ‘genetic’ processes of crossover and mutation Darwin’s natural selection uses fitnesses of organisms in their environment to improve the gene pool GA is a simple model of this process MIT Design Inquiry Genetic processes Genetic Crossover Points A B Parents Offspring C D MIT Design Inquiry Parents Offspring A A B A B C B B C B A B Crossover point MIT Design Inquiry Offspring Parents A A B A A C B B C B B B Crossover point MIT Design Inquiry How do they relate to human processes of design? l l l Can map genetic representation onto a computational representation of a design; can map phenotype onto a interpretable view of a design Humans work on single or few designs at a time/ genetics works on a population of ‘designs’ in parallel Humans can be seen to “search” design spaces - this is one interpretation of what GAs are doing. MIT Design Inquiry Why is there power in this metaphor? l l l l l l l Guaranteed improvement - Darwinian evolution Large scale search Blind search Fitness can be human evaluation Fitness can change over evolutionary time Can produce complexity Can produce unexpected results MIT Design Inquiry How can they be used to extend the capabilities of the designer? l l l Creativity through genetic engineering Novel designs through extending genetic crossover Novel designs through different fitnesses MIT Design Inquiry Genetic Engineering and Creative Genetic Design l l l Background l genes, genotype, phenotype, fitness Connecting genes to performance in fitness Emergent gene clusters ˛ evolved genes MIT Design Inquiry Total Population “bad” “good” x x x x •• •• x x “good” genotypes “bad” genotypes MIT Design Inquiry b b a rule 1 a b a a rule 2 a rule 3 a a a rule 4 b a a rule 5 a a b b b a rule 7 b rule 6 a a MIT Design Inquiry rule 8 a design 1 design 2 design 3 good good {1,12,2,8,5,4,4,2,8,5,7} {1,2,1,8,2,8,5,5,6,6,8,1} design 4 design 6 bad {6,4,1,2,8,5,4,2,8,5,3,3} {3,4,8,2,8,1,6,5,7,3} design 8 good {3,2,2,6,5,8,2,1,4,4,3,1} design 5 good design 7 bad neutral {2,3,2,3,4,3,5,6,5,1,6,2} design 9 bad design 10 neutral bad {3,1,8,5,5,6,4,6,1,1,3,3} {1,6,4,2,7,3,4,8,6,1,6,2} {6,4,1,2,3,4,5,2,1,7,4} {2,3,7,5,1,2,8,3,1,6,2,1} Composite building block A {2,8,5} MIT Design Inquiry MIT Design Inquiry MIT Design Inquiry MIT Design Inquiry MIT Design Inquiry MIT Design Inquiry MIT Design Inquiry MIT Design Inquiry MIT Design Inquiry MIT Design Inquiry MIT Design Inquiry Mondrian Genotype Form l l In form of a tree Each node has four variables direction of rectangular split (4 values) fraction of the split (15 values) colour of split area (10 values) line width (3 values) l l l l MIT Design Inquiry Fitnesses for Representation l l l l l offset between actual and required positions of dissection lines number of lines with correct line width, normalised number of correct colour panels, normalised number of lines assigned, normalised number of unassigned lines, normalised MIT Design Inquiry Genetically Engineered Mondrian Genetically Engineered Mondrian Genetically Engineered Frank Lloyd Wright Windows MIT Design Inquiry Flondrians l l l Mondrian painting ˛ genetically engineered genes: M-genes Frank Lloyd Wright windows ˛ genetically engineered genes in same representation: Fgenes “Flondrians” are the genetic product of mating M-genes with F-genes MIT Design Inquiry MIT Design Inquiry MIT Design Inquiry How Many Designs Are There and Where Are They? MIT Design Inquiry Genetic crossover as an interpolation p1 p2 g1 pc gc g2 Phenotypic space P Genotypic space G C(p1,p2) Æ pc C(g1,g2) Æ gc MIT Design Inquiry P+ p1 p2 P MIT Design Inquiry Interpolation MIT Design Inquiry MIT Design Inquiry (interactive Genetic Art III) MIT Design Inquiry Image detection (a) (b) MIT Design Inquiry Modelling Interest Reward HEDONIC VALUE 1 Hx 0 n1 n2 Nx NOVELTY -1 Punish Berlyne’s model of arousal based on novelty using Wundt curve MIT Design Inquiry Different novelty functions MIT Design Inquiry Different novelty preferences N=0 N=1 N=2 N=3 N=4 N=5 N=6 N=7 N=8 N=9 N=10 N=11 N=12 N=13 N=14 N=15 N=16 N=17 N=18 N=19 MIT Design Inquiry ...
View Full Document

This note was uploaded on 11/10/2011 for the course ARCH 4.101 taught by Professor Williamhubbardassn during the Spring '03 term at MIT.

Page1 / 36

johngerolecture - How do genetic algorithms relate to their...

This preview shows document pages 1 - 6. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online