... Good point. The competition is against other machine discovered s-expressions so it's likely they'll encounter the same limitations that we have. On the...
It turns out that supporting feed-forward networks isn't as easy as I'd hoped. I naively assumed that if I modified the 'add connection' routine to only add...
In ANJI we have an option to enable/disable recurrent connections. The way we determine if a connection is "backward" is by checking for a loop every time we...
Hi, ... Thanks, this is an option I'll keep in mind if I decide to attempt the Othello comp. Of course this approach will probably start eating CPU cycles as...
Some people here may be interested in this competition I noticed: http://eldar.mathstat.uoguelph.ca/dashlock/CEC05/EvolvedArt.html Mattias in particular maybe...
At 00:01 02/11/2005, you wrote:
>Hi,
>
>Philip Tucker wrote:
>
> >In ANJI we have an option to enable/disable recurrent connections.
> >The way we determine if...
... connections. ... checking for ... you're ... to see ... attempt the ... CPU ... prefer ... obsessed ... in the ... that ... higher ... neurons ... to ... ...
Ken, The way I did it in Semiann was to separate activation order from identity. Each node has an identity number and an activation priority. Both are...
Hi,
A-ha, you mean if I have
A -> B -> C ->D
A-> B' ->C' ->D
Then it won't let me connect C' to B?
That's right, but I thought I had a scheme which sorted...
... Yes I was thinking about this also [Actually I did a superior analysis... I used pencil and paper instead of napkins ;) ]. Does it matter I wonder? I...
At 22:05 02/11/2005, you wrote:
>Ian Badcoe wrote:
>
> >Hi,
> >
> >A-ha, you mean if I have
> >
> >A -> B -> C ->D
> >
> >A-> B' ->C' ->D
> >
> >Then it won't...
Ian, Was not able to follow your "bit encoding" notion (flying through things, will re-read later when I can concentrate). ... individual 2 has just connected...
Hi all, I'm evolving an extension of Stanley & Miikulainen's robot duel domain in ANJI 2.01, but I'm having trouble setting the speciation threshold. If I set...
Hi, If you look back a few weeks to me posts on "continuous speciation control" that might help you. I was doing things differently in that I was adding and ...
Hi, Right after I posted my question I realized I had just violated the first law when posting in forums/groups: before asking, always search back to see if...
Hi, The large variation inteh distribution of genomes among species is something that is expected if the fitness of the species also varies a lot - the fittest...
... Ok yeh I've got it. Each node has a bit array that indicates which nodes (through however long a pathway) connect into it. I suppose you could do it the...
My initial function approximation experiment tried to approximate y=f(x) by feeding values of x and comparing an ANN's output with the expected output. So the...
I don't know much about signal processing, but would it be possible to have a dynamic sampling rate, where an output of the network controls the speed, and...
Hi, Well, I think I finally got it nailed down. I've been tinkering with Ian's method all morning trying to get the right parameters and conditions for raising...
At 20:56 06/11/2005, you wrote:
>Ian Badcoe wrote:
>
> >One cheap way to handle this might be to keep a bit-array on each
> >node. Each node owns a unique bit...
At 03:54 04/11/2005, you wrote:
>Ian,
>
>Was not able to follow your "bit encoding" notion (flying through
>things, will re-read later when I can concentrate)....
At 11:59 06/11/2005, you wrote:
>Hi,
>
>Right after I posted my question I realized I had just violated the
>first law when posting in forums/groups: before...
... Isn't that handled by the assignment of number of offspring to a species? If species X is sceduled for extinction, then the number of offspring that it is...
... Hi Bas, I thought I'd chime in and back the conversation up a little from implementation to theory. Why are we using speciation (and along with it explicit...
At 18:29 07/11/2005, you wrote:
>On 11/7/05, Ian Badcoe <ian_badcoe@...> wrote:
> > the ideas behind the second of these are quite complex, and will...
... Perhaps. Recognize speciation for what it was: an attempt to protect genetic diversity. Explore the possibility that speciation, while being *A* way to...
Having problems with message search?
Fill out this form to ensure your group is one of the first to be migrated to the new message search system.
