Hi all, Could anyone explain the working mechanism of Stochastic Universal Selection Technique Thanks in advance :)

This thread is the first hit on Google for that topic... So anyone who replies with a plausible explanation will create the first definition :-)

Hehe.


On a more serious note, the second hit in Google is a link to an extract from Mat Buckland's first book, which explains it exactly. Also, the third hit on Google is Wikipedia, which explains it pretty well also...

Alex

P.S. Do you search before asking questions?

@alexjc
Thanks Alex for your reply. Yes I've searched but the problem that I've read the definition of this technique from 3 sources:
- AI Game Engine Programming "Brain"
- AI Techniques for Game Developers "Mat & Andre"
- Wikipedia

The three sources "from my understanding" explains this technique in different ways so I asked here to resolve that conflict

Why not post your understanding of the parts that conflict rather than ask for the whole definition again -- which may result in more conflicts?

@AlexJC: Thanks for reply
My understandings:-
> This step is common between all:
Roulette while is constructed where the fitter individual have bigger slice
- AI Game Engine Programming:
To choose offspring:-
1-Don't spin the roulette
2-Take the number of individuals you want to select
3-Select the owner of the roulette slice at 1/n increments along the wheel (Can't understand anything from this step!)

- AI Techniques:
To choose offspring:-
1-Don't spin the roulette
2-Put Pointers in the middle of the roulette and pick the pointed slices

- Wiki:
SUS uses a single random value to sample all of the solutions by choosing them at evenly spaced intervals (Can't understand anything from this step!)

So what the real definition of Stochastic Universal Selection :D

Thanks Alex :)

Looks like they're all doing the same thing, just implemented differently. Because you don't understand the details of the implementation of some books, it may seem like they're doing something different but it's rather straightforward.

This is all you need to know:
"The fitter individual are selected proportionally to their fitness."

How would you implement that?

@AlexJC: Thanks for reply