 |
|
|
 |
|
|
|
|
witeflame |
| | | | | | posted 6/29/2010 11:52 |   | | | | | | | | | Good afternoon (from here on an island...), I have this exciting Thesis, all about a Hybrid AI system. I have to make sure that it can have an 'Anytime-algorithm' ability.
Any Ideas? Any suggestions will be greatly appriciated.
*Its a Genetic algorithm path planning hybrid
| | |
| | |
hunt |
| | | | | | posted 6/29/2010 18:38 | | | | | | | | | | What is an "anytime algorithm"?
Edit: Ah, nevermind. Google provides, as always. So from the definitions I've read, it's basically an algorithm that converges quickly to within acceptable error, though perhaps takes a while to fully optimize.
| | | Last edited by hunt @ 6/29/2010 6:43:00 PM | | |
| | |
witeflame |
| | | | | | posted 6/29/2010 21:02 | | | | | | | | | | yup, thats right.
It basically gives an output regardless of the computational time required, however as the computational time increases, so does the quality of the output.
The output is however always valid.
| | |
| | |
tkorrovi |
| | | | | | posted 6/30/2010 16:42 | | | | | | | | | | Yeah, different ways to emulate that which naturally occurs in self-developing systems. No, i don't say that it is wrong, any "natural" principles are useful. Apply the ouroboros principle, too ;)
| | |
|
|
|
|
|
