Background notes from 09/16/09
September 23, 2009by Tonje Stolpestad (tstolpes)
-by Tonje
Genetic Programming, Push, PushGP
Genetic Programming for finite Algebras
GECCO -publishing?
Finite Algebra
Boolean algebra
Primal every possible operation can be expressed by a term using only AND, OR and NOT
All Boolean functions can be made of NAND
Number of terms grow astronomically with size of underlying set
Find terms with certain properties
Discriminator term, determine primality
Malcev majority and pixley terms
Genetic Programming
Random Generation
-> Assessment
->Selection
->Variation
->Assessment
->Solution
Tournament selection
Elitism not so a good thing
Enough variation
Koza http://www.genetic-programming.com/johnkoza.html
Genetic Hitchhiking
The Push3 Execution Stack and the Evolution Control
Push v3
Open ended evolution of arbitrary computational process
Natural selection or programs expressed in syntactically trivial, semantically rich program representation.
Rich enough to express evolving programs of arbitrary architecture and even the mechanisms of evolution itself
Stack based postfix language
Trivial syntax
Evolved programs may use
Multiple data types
Subroutines
Recursion and iteration
Supports evolution of evolutionary mechanisms
Parentheses are critical for code manipulation
Dynamic
First step
Create Blog
Read what wolf and understand
Diversification of mitochondrial dna
Show what you can and can not know
Code
Geography
Panmitic population
Charles Ross
Meet
What would the biological community learn
What is wrong with the three of life
Inter fertile
Color web
Trivial Geography in Genetic Programming
Tag-Mediated Altruism
Tags and Tolerances
Varied interaction radius within a linearly structured population
Tags: background, notes