Presentation is loading. Please wait.

Presentation is loading. Please wait.

CS 447 Advanced Topics in Artificial Intelligence Fall 2002.

Similar presentations


Presentation on theme: "CS 447 Advanced Topics in Artificial Intelligence Fall 2002."— Presentation transcript:

1 CS 447 Advanced Topics in Artificial Intelligence Fall 2002

2 Administrative Class roll Syllabus Office hours Lecture exam date Course form

3 Evolution Individuals Population Environment Fitness Selection - selective pressure Reproduction Competition – survival of the fittest

4 Heredity Asexual versus sexual reproduction Genes Loci Alleles Genotype versus phenotype Genetic operators: replication, recombination, mutation

5 Evolutionary Computation Solving “difficult” problems Search spaces: representation & size Evaluation of trial solutions: fitness function Exploration versus exploitation Selective pressure rate Premature convergence

6 EnvironmentProblem (search space) FitnessFitness function PopulationSet IndividualDatastructure GenesElements AllelesDatatype

7 Evolutionary cycle selection reproduction mutation competition evaluation initialization

8 Pros General purpose: minimal knowledge required Ability to solve “difficult” problems Solution availability Robustness Cons Fitness function and genetic operators often not obvious Premature convergence Computationally intensive Difficult parameter optimization

9 Math problem Given the function f(x,y) = x 2 y + 5xy – 3xy 2 for what integer values of x and y is f(x,y) minimal?

10 Solution space: Z x Z Trial solution: (x,y) Gene representation: integer Gene initialization: random Fitness function: -f(x,y) Population size: 4 Number of offspring: 2 Selection: exponential Genetic operators: 1-point crossover Mutation (-1,0,1) Competition: remove the two individuals with the lowest fitness value

11 Problem solving steps Collect problem knowledge Choose gene representation (e.g., bits, integers, floats, parse trees, musical notes, …) Design fitness function Creation of initial population (e.g., random, manual, based on previous run, …) Parent selection Decide on genetic operators Competition / survival Choose termination condition Find good parameter values

12 Parameters Population size Selective pressure Number of offspring Recombination chance Mutation chance Mutation rate

13 procedure EA begin t  0 initialize P(t) evaluate P(t) while (not termination cond) do begin t  t + 1 select P(t) from P(t-1) alter P(t) evaluate P(t) end

14 History of EC Genetic Algorithm (GA) John Holland 1960’s & 1970’s fixed-length binary strings binary crossover & mutation Evolutionary Programming (EP) Lawrence Fogel 1960’s finite state automata exclusive use of mutation

15 Evolution Strategy (ES) Ingo Rechenberg & Hans-Paul Schwefel 1963 real valued function optimization employing multi-variate zero-mean Gaussian mutation Genetic Programming (GP) John Koza 1990 individual = computer program in the form of a parse tree (typically LISP is used) recombination most important genetic operator

16 Evolving versus learning In EC learning occurs at the population level instead of at the individual level In nature evolution and learning are combined Darwinian evolution evolves the blue print of a learning system Lamarckian evolution involves direct inheritance of characteristics acquired by individuals during their lifetime Baldwin effect: phenotypic plasticity (e.g., learning [local search])


Download ppt "CS 447 Advanced Topics in Artificial Intelligence Fall 2002."

Similar presentations


Ads by Google