|Wesley R. Elsberry
Joined: May 2002
|Quote (GaryGaulin @ Nov. 23 2012,18:44)|
|Quote (k.e.. @ Nov. 23 2012,10:38)|
|Quote (damitall @ Nov. 23 2012,18:11)|
|Quote (GaryGaulin @ Nov. 23 2012,09:26)|
|Quote (k.e.. @ Nov. 23 2012,08:25)|
|keep googling troll|
I was refreshing my memory. Still did not run across the list of papers pertaining to molecular intelligence, but found this one:
Solving Traveling Salesman Problems Using Molecular Programming
Note: Solving a Traveling Salesman Problem is a "programming" challenge, not an indicator of "intelligence".
What, you mean like making virtual critters move towards targets on a monitor screen?
He's the classic fundy marksman.
Shoots at a barn wall then goes over and draws a bullseye on the hole.
That paper has nothing to do with 'molecular' intelligence or much to do with the TSP.
They are using a GA for small scale analysis of hydrogen bonds in DNA because the TSP looks a bit like the math involved in the bonds.
Something I'm pretty sure Gary doesn't understand.
Gary please explain how the TSP relates to hydrogen bonds.
I was speaking in reference to this argument that began on page 11 where the traveling salesman problem was presented as a good indicator of which model is a more accurate representation of reality (where living things each have their own intelligence):
As usual, Gary gets it wrong. Back on page 11, I asked Gary to substantiate his claim. What claim, you might ask? Let's review:
|Quote (Wesley R. Elsberry @ Nov. 07 2012,00:48)|
It's more technologically demanding, but as I earlier mentioned the model puts EA's and GA's to shame, as though they are baby-toys.
To make that claim stick, you'd have to have done a broad survey of capabilities of evolutionary computation instances and made specific comparisons demonstrating the superiority of your approach. I doubt that this has happened. You could dispel that by showing your work. Let's start with your comparison of your program and that of Eureqa. Please show us how your program does symbolic regression better than Eureqa. And then PyEvolve. And DEAP. And PyGP. Your claim implies that you've already taken this step, so all I'm asking for is that you show us what you must already have in hand.
Others have already asked for a similar comparison concerning the TSP. I'd be interested in that, too. I asked Bill Dembski to make his criticism of GAs stick when considering the TSP back in 1997, and so far as I know, he has never even attempted a discussion in general that focuses on the TSP. Can you do better?
The claim was that Gary's work outperforms evolutionary computation approaches. This is a quantitative claim. I asked to see the benchmarks that Gary had indicated that he had already performed in order to come to this conclusion, benchmarks being a necessary prerequisite to even make the claim.
Gary's backtracking seems to be along the lines of saying that his program and evolutionary computation don't address the same concerns, in which case there is no basis to say that anything at all that isn't addressed to the same concern is like "baby-toys".
Gary could clear this up in a moment. If a technology doesn't address the concerns that his program does, then he can just say that it isn't relevant to what he is doing, and retract any false claims that he has made a comparison or even could make a relevant comparison.
On the other hand, if Gary wants to insist that there is some aspect on which his program and evolutionary computation is comparable, he needs to pony up more than an assertion that he has something superior. And it needs to involve more than an assertion that the compared technology isn't aimed at accomplishing Gary's personal mission. That leads to a conclusion that Gary is making a category error, not an informed assessment of capability.
"You can't teach an old dogma new tricks." - Dorothy Parker