<I apologies if you recevied the multiple copies of the CFP> SECOND CALL FOR PAPERS KBCS-2004 THE FIFTH INTERNATIONAL CONFERENCE ON KNOWLEDGE BASED COMPUTER...
Very cool visualization interface! I noticed a few minor things though: -I don't think the default value for connectness of the initial genome should be 0.2....
Hi Derek, Ken, ... You're right that culling stagnant species does often give rise to 'jumps' in a search. But do you agree that some gradual improvements come...
... Thats right Chad. Certainly before the days of dedicated built-in FPUs floating point math was painfully slow. But even in todays CPUs with highly...
... I take the point about performance of the NEAT code. I tried measuring the my gene comparison/compatibility code a while agoand it was not measurable for a...
... I am quickly becoming convinced that this is not nessearily sesnsible to do all at once. It makes sense to decompose the behaviours for one network to...
... Here's a thought... Why give connections a unique innovation number at all? Why not assign their unique ID based on the innovation numbers of the nodes...
... between ... are ... performance ... measuring ... not ... innovations ... approach ... problem ... NEAT ... matching ... and ... Someone once pointed out...
... innovations ... approach ... problem ... NEAT ... matching ... and ... can. ... number at all? ... of the ... 253 and ... duplicates. ... connects ... ...
... I think there is a problem to appear. You idea supposes some global pool for connections information. Right? So there is no way to keep connections that do...
... I originally thought of one hash per organism. But now I think it would be one hash per node. Too much work in C/C++, but very easy in Perl... $network =...
... I think I can live with that. So...I've actually started coding. YEAH!!! And I have some questions. For the initial state of the experiment, it is a...
In the case of natural neural networks, do we have more than one connections between two neurons? If so, should we or shouldn't we incorporate them in ANNs ? ...
Natural neural networks are hardware-limited. They can't represent a weight more than some limit, by virtue of their physical structure. Having multiple...
Since, natural neural networks work good enough :), couldn't decreasing our precision and allowing multiple connections be a better way of making ANNs? I dont...
Hi all, I just took the GRE last weekend, with an eye toward applying for graduate school. I don't have a strong CS background, and I'm not really interested...
Well, I graduated from UT/austin and am now a master's student here at the University of Birmingham (UK), studying Natural Computation. It really sounds...
On Wed, 2 Jun 2004 17:08:50 +0100 (BST), Kennon Lynn Ballou ... Thanks for the link. It does sound like what I'm looking for...though I was hoping to find a...
... Yeah the Sussex program is supposed to be pretty good. I've actually been talking with Phil Husbands off and on, and he seems like a really great guy. That...
... From what I saw, most US universities at least require you to do basically the same amount of work as a master's degree before actually fully launching on...
... That was just one possible explanation for fitness jumps following stagnant stretches. There may be others. ... I'm not really sure that is the question....
... I don't know of any experiments on this with NEAT. XOR's are not necessarily good benchmarks though...they tend to have very unusual characteristics. ken...
... Agreed. ... I'm still not sure if you've grasped my main point here. Ok so sometimes a new connection appears and it immediately increases the fitness of a...
Hi Umer, I agree with John that multiple connections between two given neurons would be redundant. You can still represent the full range of values AND reduce...
... I kind of like the idea. As a substitute for the loss of history I was thinking of storing the generation number at which a connection was created. Anyway...
... sometimes ... of a ... make ... Colin, I pretty much agree with your main point, but this point is technically not correct. Adding a node *does* slightly...
Guys one thing worth clarifying is that expressed DNA that does not change the phenotype is not the same as useless appendages hanging around. What's...
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.
