블로그 이미지
Leeway is... the freedom that someone has to take the action they want to or to change their plans.
maetel

Notice

Recent Post

Recent Comment

Recent Trackback

Archive

calendar

1 2 3 4 5 6
7 8 9 10 11 12 13
14 15 16 17 18 19 20
21 22 23 24 25 26 27
28 29 30
  • total
  • today
  • yesterday

Category

2008. 3. 27. 17:25 @GSMC/정문열: Generative Art
2008-03-27 나무 @504

(1) Tree of Exp
(2) Randomization ( tree의 leaf들을 random하게 selection한다.)
(3) Mutation & Cross-over


randomization

uniform distribution 균등 분포
: 각각이 선택될 확률이 동일
http://en.wikipedia.org/wiki/Uniform_distribution_%28discrete%29

ref. http://en.wikipedia.org/wiki/Probability_distribution


standard normal distribution
http://en.wikipedia.org/wiki/Standard_distribution
(그림)
각각의 op가 선택될 확률은 다르나, 모두 합치면 1이 된다.

(각 op가 뽑힐 확률은 지정해 둔다.)
selection을 (충분히 많이) 무한 번 반복하면 각 op가 가지는 확률을 알 수 있다.

tree 1과 tree 2를 교배시켜 새로운 tree 3을 생성할 때,
각 tree 1과 tree 2의 요소(node)가 변이(mutation)을 일으킬 수 있다.



mutation

tree의 각 node에서 mutation이 일어날 확률을(여부를) 계산(선택)한다.
즉, 각 node마다 mutation probability를 가진다.
(그림)



cross-over

(그림)

(1/2)*(1/2) = 1/4
그러나 (유전자 쌍의 동일한 위치에서 일어나는) cross-over 때문에 매번 다른 유전자를 지닌 자식이 나온다. => "phenotype"
대응되는 node끼리 교배 -> 우열 등의 규칙은 임의로 결정

(그림)


tree 하나를 염색체 하나로 보고 유전 법칙을 적용하자.
염색체가 (이중나선 DNA)쌍으로 존재하듯, 염색체 한 쌍을 보유하는 개체 하나 당 tree 2개로 구성되어 있다고 본다.
교배는 이 개체와 다른 개체, 즉 총 4개의 tree가 mutation, cross-over 수반한 교배를 한다.
이렇게 생성된 evolved 이미지에 criteria를 적용하여 적자생존의 법칙처럼 적당한 이미지가 걸러지도록 한다.



"random 함수는 동일한 확률 분효를 가진다."

(그림)



'@GSMC > 정문열: Generative Art' 카테고리의 다른 글

Class와 Object  (0) 2008.04.14
[이선애] study 1  (0) 2008.03.28
tree (data structure)  (0) 2008.03.24
class 3/20  (0) 2008.03.20
references for basic programming  (0) 2008.03.17
posted by maetel