Monday, June 04, 2007
Estimating a Cell's Contribution to Directed Mutation
There is a great presentation available from the International Symposium on Neural Networks on the No Free Lunch algorithms. The presentation points out that "evolutionary algorithms" often implicitly import solution space structure information into their search algorithms. It even gives an example of how to measure the imported information in bits based on the difference between the performance of the search algorithm and a random search. The presenter mentions that including the number of bits imported into the search algorithm should be included in every evolutionary paper. However, I think that there is a more significant idea lurking in there somewhere.
Since, according to Marks, you can measure the number of bits imported into a search algorithm by running trials, why don't we apply this to known cases of evolution? It seems that we should be able to calculate the number of bits the cell itself is applying to its own evolution, shouldn't we?
Now, before getting too excited, there are some caveats. The biggest one is that Marks doesn't separate importing information into the search algorithm versus importing information into the environment. Just like you can import bits into the search algorithm, given any search algorithm, you can also import bits into the environment to match the algorithm. So is the design in the search or in the environment? It's a thorny issue.
In any case, even if we can't follow all of the possibilities, we should at least be able to get an estimate of the number of bits being supplied by the cell for its own evolution. This would be interesting to apply to some of the existing data for evolutionary scenarios.
Since, according to Marks, you can measure the number of bits imported into a search algorithm by running trials, why don't we apply this to known cases of evolution? It seems that we should be able to calculate the number of bits the cell itself is applying to its own evolution, shouldn't we?
Now, before getting too excited, there are some caveats. The biggest one is that Marks doesn't separate importing information into the search algorithm versus importing information into the environment. Just like you can import bits into the search algorithm, given any search algorithm, you can also import bits into the environment to match the algorithm. So is the design in the search or in the environment? It's a thorny issue.
In any case, even if we can't follow all of the possibilities, we should at least be able to get an estimate of the number of bits being supplied by the cell for its own evolution. This would be interesting to apply to some of the existing data for evolutionary scenarios.
Permanent Link posted by crevo @ 9:21 PM 
Comments:
<< Home
Your first comment was kind of simplistic. Care to expand? I didn't understand the second comment. Care to explain?
I am curious, why do you need to "estimate a cell's contribution to direct mutation"? You already know the results of all experimentation, and, besides, if God decides to change how things are done, won't that render even your expected results moot? So, why bother?
# posted by 
: 6:50 PM
: 6:50 PM
'I am curious, why do you need to "estimate a cell's contribution to direct mutation"?'
There are several goals. On a personal basis, I love understanding new things, and understanding how God worked the genome seems to be the pinnacle of interest. The stuff I've learned so far is really cool, especially the ways in which it seems God applied metaprogramming (program-generating-programs) into the immune system.
On a more practical note, learning what causes change from generation to generation will help learn the direct causes of genetic disease, and will help future transgenic technologies understand how to better play nicely with the genome.
"You already know the results of all experimentation"
I do? Perhaps I have merely forgotten.... Care to explain your thoughts?
"if God decides to change how things are done, won't that render even your expected results moot?"
Certainly this is true. There are a lot of things that God has, could, and will do that will render a great many things moot. Even if you didn't believe in God, you would have the same conundrum -- life throws a great many unexpected things at you that can render everything you've done moot. But we are here, and here is where we should live. In the meantime, I enjoy learning more and more about God's creation.
Post a Comment
There are several goals. On a personal basis, I love understanding new things, and understanding how God worked the genome seems to be the pinnacle of interest. The stuff I've learned so far is really cool, especially the ways in which it seems God applied metaprogramming (program-generating-programs) into the immune system.
On a more practical note, learning what causes change from generation to generation will help learn the direct causes of genetic disease, and will help future transgenic technologies understand how to better play nicely with the genome.
"You already know the results of all experimentation"
I do? Perhaps I have merely forgotten.... Care to explain your thoughts?
"if God decides to change how things are done, won't that render even your expected results moot?"
Certainly this is true. There are a lot of things that God has, could, and will do that will render a great many things moot. Even if you didn't believe in God, you would have the same conundrum -- life throws a great many unexpected things at you that can render everything you've done moot. But we are here, and here is where we should live. In the meantime, I enjoy learning more and more about God's creation.
<< Home
