COMMENTS
All posted comments will show up on this page instantly. Except for anything with http://, which I've disallowed to prevent spam.
_____________________________________________________________________________________________________
Test.
_____________________________________________________________________________________________________
Hi! I understand what you are looking for, but GP may not be the best method to attack your problem. Can you let us know what the disease is? This would give us a starting place in determining if evolutionary based algorithms are even close to appropriate. I posted an entry at my blog -> http://www.deepmarket.com/machine-learning/meta-genetic-programming-to-help-cure-a-disease/
_____________________________________________________________________________________________________
If you want to look into biology you might be interested in Aubrey de Grey, who looks upon death as something that needs to be fixed: and he's the one to do it.
http://www.technologyreview.com/articles/05/02/issue/feature_aging.asp
is a Technology Review article by Sherwin Nuland, author of the book "How We Die," who does not share de Grey's views on death but gives a simple account of his ideas. Page 3 of the 7 page article lists de Grey's 7 causes of death and his ideas for eventually dealing with each. That page also has a link to de Grey's web site.
_____________________________________________________________________________________________________
As you say on the FAQ page, you ultimately want to use computers to cure disease. Are you familiar with the method of splitting up a problem and having many people's computers working on it? (Called Distributed Computing.) SETI@Home is the most famous example, although it is not a medical system. Look at this page of projects:
http://distributedcomputing.info/projects.html
and particularly the column marked "Life Sciences." I don't know anything about setting up something like this.
Also, you can keep up to date on tests of new drugs at:
http://www.clinicaltrials.gov/
_____________________________________________________________________________________________________
Free artificial-intelligence software made with Java:
_____________________________________________________________________________________________________
You mentioned in part E. of the Meta Genetic Programming that one other way was to concentrate on one section.
I believe that is a more organized method and if done properly can save an enormous amount of time compared to a method of tackling the entire job at once.
_____________________________________________________________________________________________________
Continuation:
If you use the more concentrated section method, and if you happen to also know which part of the entire system is the slowest then should probably work on that section firt.
The reason is: The total speed of a system will always be dependen on its slowest part (i.e. limited by the slowest part).
So by working on the slowest part first you can efficiently increase the rate of your system substantially.
____________________________________________________________________________________________________
You sound exactly like me, haha. http://www.editthis.info/creation_engine/Main_Page My email is kiloiscrazy@hotmail.com
_____________________________________________________________________________________________________
Looks like the same bot that tryed several times to hace my email form. it just susceeded on sending a jiberish mail to 8 ppl via cc on it's 12 atempth :(
I guess i should prevent any impout contaning mime looking date.
_____________________________________________________________________________________________________
I've tried sending spam emails through the form and it doesn't work. Apparently the problem affects a different type of CGI script. Someday I will get around to deleting all the junk posts.
--Michael Gospatrick
_____________________________________________________________________________________________________
GP is still quite "young" in computer science (otherwise it's use would be more prolific)... which means meta-GP which is little understood as yet, will still take at least number of years before it starts being applied (to things like finding cures for diseases).
Often employing the better understood technology / sciences and making incremental improvements are the fastest way when you really need to get a solution within a timeframe rather than researching for a better solution who's fruition cannot be time-bound.
Despite my previous advice, I love the concept of meta-gp and hope you don't give up on it.
__________________________________________________________________________________________________
I think the fundamental problem with genetic programming is the weakness of the design. You can have a program that writes different, evolving programs to solve a problem, simple or complex. However, if the efficiency of the program-writing program is not optimized, then it will take a long time to produce results. There are a multitude of different ways to approach the problem. You could have it rewrite itself and run a fitness function to determine its effectiveness/efficiency. But what if the fitness function is inefficient itself? I think that you would need a system that is organized in a particular hierarchy. A master program that writes programs. It even copies itself to rewrite itself in parts or modules. Also, it has to be able to evaluate the fitness of the fitness function. Of course this is what is the only problem. Stacking fitness checks might cause a infinite fitness loop. Removing people from the programming is what is ultimately ideal and It needs a system of checks that is autonomous.
_____________________________________________________________________________________________________
Hellojjp - this is just a testing, don't worry about it
_____________________________________________________________________________________________________