1 / 17

Spring 2011 Artificial Intelligence COSC 40503

Antonio Sanchez Texas Christian University. Spring 2011 Artificial Intelligence COSC 40503. References www.cs.vu.nl/~gusz/ecbook/slides/Genetic_Algorithms.ppt A.E. Eiben and J.E. Smith, Introduction to Evolutionary Computing Genetic Algorithms web.umr.edu/~ercal/387/slides/GATutorial.ppt

otylia
Download Presentation

Spring 2011 Artificial Intelligence COSC 40503

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Antonio Sanchez Texas Christian University Spring 2011Artificial Intelligence COSC 40503 References www.cs.vu.nl/~gusz/ecbook/slides/Genetic_Algorithms.ppt A.E. Eiben and J.E. Smith, Introduction to Evolutionary Computing Genetic Algorithms web.umr.edu/~ercal/387/slides/GATutorial.ppt igda-phx.org/resources/20031121_ga.ppt parasol.tamu.edu/~amato/Courses/ 689-608/presentations/jkp2866/gep-jack.ppt www.niss.org/affiliates/proteomics200303/ presentations20030306/04%20Jennifer.ppt http://www.eece.unm.edu/~akbazar/

  2. A formal presentation of GA’s“The Gene is by far the most sophisticated program around.”- Bill Gates, Business Week, June 27, 1994 When to use them • Alternate solutions are too slow or overly complicated • Need an exploratory tool to examine new approaches • Problem is similar to one that has already been successfully solved by using a GA • A quasioptimal solution is allowed When no to use them • If you need the optimal solution • If you have the Analytical solution or it can be easily solved (one variable functions) • If the solution is a convex function • If the space is discrete and enumerable

  3. Lamarckian Evolution • Lamarckian Theory • Based on the concept of use and disuse • Over a few generations, a given structure or organ will increase in size if the creature and its parents use that structure often. • On the other hand, if a structure and organ is in disuse it will get smaller and even disappear in subsequent generations.

  4. An Example of Lamarckian Evolution • A Giraffe has a long neck because its ancestors used its neck to reach food. • Based on Lamarck’s theory, the Giraffe of the future will have an even longer neck than its contemporary relatives.

  5. Darwinian Evolution… • All animals are constantly changing and evolving • The primary goal of an animal is to mate and have as many offspring as possible • Concept of natural/sexual selection • Natural selection, development, and evolution requires time

  6. Genetic Operators • Encoding Method of representation of the coefficient in a gene Note that if there is no encoding then we talk about evolutionary programming • Mutation When a bit inside the chromosome is invested (extremely rare) • Internal positional swap When two bit inside the chromosome exchanges their position (arity=1) • Reproduction When a new population is generated • Crossover (aka recombination) When randomly two chromosomes are chosen bits are crossed over generating two new individuals (arity = 2) • Substitution When the parents of a crossover are eliminated from the population • Decoding Process of obtaining of the value for a coefficient

  7. GA general process • Determine the task to evaluate • Codify the Chromosome and number of genes • Define the initial population • Criteria of selection by the environment • Random selection of survivors • Chromosome crossing • Mutation • Evolution

  8. 1/6 = 17% B fitness(A) = 3 A C fitness(B) = 1 2/6 = 33% 3/6 = 50% fitness(C) = 2 GA operators: Selection • Main idea: better individuals get higher chance • Chances proportional to fitness • Implementation: roulette wheel technique • Assign to each individual a part of the roulette wheel • Spin the wheel n times to select n individuals

  9. Initial Population Although the initial population is assumed random you may consider the following options • Intelligent Design: If possible start with some range values for the function coefficients to faster convergence. However remember that this might work like the Crash Cut of Minimax • A Priori Seed: Similarly you can start with some given alleles or seed values. For example use the values you obtain from your own experience. • Diversity control to avoid Inbreading and Locking If you use the previous considerations remember to add some diversity either by mutation or in the evaluation function to allow to step out of the box you have predefined

  10. Phenotype/Genotype Mapping The mapping can be • Encoded such is the case of classical GA • Uncoded or direct such is the case of EA or Evolutionary Algorithms And so the genotype can represent a • An Integer • A Real number within a range • An assign task within an scheduling problem • The branch of a decision tree (an S expression) • A case within a enumerable switch

  11. Best in the generation Generation performance Chromosomes in time To determine which chromosomes are selected in next generation may include aspects such as: • Diversity • Parallelism • Elitism • Specialization Niches • Meta Chromosomes • Substitution rather than Addition • Deception • Always plots • Best in generation • Generation as a whole

  12. On Natural Genetics • ALL the proteins of the live beings in our planet are composed of sequences of 20 amino acid • The DNA that forms the chromosomes is built by four nucleotidos in the shape of double helix: • Purines: Adenosine, Guanine • Pyrimidines: Thiamine (or Uracil), Citosine • The code is TEtranary digiT (TET) = {A,G,T,C} • Codons are triplets of nucleotides (ATC, GTC,…) just as bytes related to 8 bits. Yet there are only 20 aminoacids so there is great amount of redundancy in the codes. • Furthermore the DNA contains a high percentage of chains that are not used, yet help in schemata resolutions. • The DNA is used in the three living species: • Plants, Animals and Fungi

  13. Schemata Analysis • Schemata are similar strings of bits that contain either {0,1,*}. For example • *0000 = { strings of 5 bits that begin with a 0 or a 1 = { (10000) , (00000) } • *00* = { strings of 4 bits that begin with a 0 or a 1 and end with a 0 or 1 } = { (0000, 0001, 1000, 1001) } • Thus every population of chromosomes can be typified diverse schemata; specifically the population is going to have (L+1)^k schemata where k is the cardinality of the language (in this case 2) and L is the length of the chromosome. • Every schemata H has some properties such as • Order OR (H) : I number of fixed positions. • Examples: Or (*1*1*1*1) = 4, OR (**11*1) = 3 • Length dL (H): Greater distance among the fixed positions of H • Examples: dL (*1*1*1)= 5-1 = 4, dL (*111*)= 4-2 = 2

  14. Schemata Analysis • It is intuitive to notice that the DL of a given H plays an important part in the probability of destruction or survival of the H through the time. • The longer the string that defines H the greater its probability of extinction in the next generations • For this reason an important hypothesis of GA’s is that the construction blocks of a genes in a chromosome should be small • The fundamental Theorem of the AG tells us that: • Schemata of low order, with reasonable adaptation, will have an exponential growth in the number of its instances in subsequent generations of chromosomes

  15. Softcomputing Softcomputing refers to a set of problem solving methods used when the required solution • Permits some degree of imprecision • It is subject to uncertainty • Deals with knowledge(beliefs, choices) rather than information(truth) • A quasi optimized solution is acceptable • Allows ample time to come with the solution In this case soft computing such as GA’s offers solutions with • Robust results • Low cost • Easy tractability

  16. Sources and References * Mohammad Akbarzadeh a tutorial on Genetic Algorthms URL: http://www.eece.unm.edu/~akbazar/ * Sanghamitra Bandyopadhyay URL: http://ranger.uta.edu/~bandyopa/demo_uta.ppt * Tracey Cool URL: http://www.msm.cam.ac.uk/phase-trans/2002/Harrypresarial/ * URL: www.csug.rochester.edu/users/ugrads/tkollar/ content/csc240/RoboCupLecture2a.ppt * AE Eiben and JE Smith URL: www.cems.uwe.ac.uk/~jsmith/ecbook/ slides/Genetic_Algorithms.ppt * Zbigniew Michalewicz URL: http://www.cs.itu.edu.tr/~etaner/courses/BE_EvoComp03_04/ * John Siirola URL: giskard.cheme.cmu.edu/class/GeneticLecture/ 11-genetic-algorithms.ppt * Wendy Williams URL: http://web.umr.edu/~ercal/387/slides/GATutorial.ppt * Trevor Meyerowitz & Claudio Pinello URL:http://www.dac.com/40th/40acceptedpapers.nsf/0/713e4a6002949a7f87256dc60058c222/ $FILE/dp19_3.PPT

  17. Sources and References * Thais Melo URL : http://www.cs.princeton.edu/~tmelo/pres.ppt * Gerardo Mendoza and Dan Reich URL: http://cow.math.temple.edu/~cow/cgi-bin/manager * Jim Cohoon and Kimberly Hanks URL: http://www.cs.virginia.edu/~evans/bio/slides/ga1.ppt * Jennifer Pittman URL: http://www.niss.org/affiliates/proteomics200303/ presentations20030306/04%20Jennifer.ppt * Peter Cowling, Graham Kendall, Limin Han + URL: http://www.cs.nott.ac.uk/~lxh/cec2.pdf * Eyal Allweil and Ami Blonder URL: http://www.cs.huji.ac.il/course/2002/aisemin/talks/week8a.ppt * Noyan Turkkan URL: http://www.umoncton.ca/turk/Genetik201.xls * EfraimTurban & Jay E. Aronson URL: http://mgtclass.mgt.unm.edu/Bose/MGT%20539/ * Hitch Hiker URL: http://www.cs.bham.ac.uk/Mirrors/ftp.de.uu.net/EC/clife/www/Q99_E.htm * Jim Smith, Agoston Eiben & Jano van Hemert http://www.evonet.polytechnique.fr/evoweb/resources/flying_circus/slides/

More Related