Hi Jeff, Did you notice that Caddie is also a part of the new gmail autopilot feature? Check it out here: http://mail.google.com/mail/help/autopilot/index.html...
1st April anybody ? Have you guys seen the google home page today? http://www.google. com/ Look for the CADIE link: If it is no longer on the home page, here...
Hum, or maybe the association of "evolutionary computation", "neural networks" and a few other related keywords in the same sentence tend to trigger the...
As Jack Tramiel from Commodore said way back in the '80s (paraphrased), "I don't care what the press says about me, good or bad. As long as they are saying...
Exactly. I think it's pretty cool that for their April Fool's joke they chose neuroevolution as the tech that might produce the first sentient AI program. :-) ...
"Cornell University computer scientists Michael Schmidt and Hod Lipson devised an algorithm that will deduce laws about the motion of a nonlinear dynamical...
Amazing what you can do with some symbolic regression and good publicity skills. The technique's been around for some 20 years, after all. Not that I in any...
Our new paper, which will appear at GECCO this summer, explores the idea that novelty search is a potentially effective tool for evolving neural networks that...
Hi all, I'm posting this link because although it's not about NEAT I think it's sufficiently interesting to anyone doing work with neural nets to warrant a...
Colin, thanks for the link. Actually, I'd seen this before and it is interesting. However, I still think something significant is missing: Like almost any...
But the only reason they're now incomprehensible scrambles of dots from a human perspective is because you and I haven't grown up with the concept of those...
Thomas, that's an interesting hypothesis that we would have no trouble with arbitrary dot patterns if we had grown up with them. That hypothesis can't be...
Oops, I think yahoo compressed my nicely scrambled dot pattern example in some browsers, making it looks like just a square of dots :( Please take my word for...
(I don't post here often, but this talk has been eye-opening.) Hi Ken. I understand your concern. It's certainly a valid one. (I could be totally wrong. I wish...
JT, my feeling is that there is a critical difference between a network that comes to represent geometry by learning about it and one that "sees" it from the...
Hello all- Below is a link to a new HyperNEAT paper that I will present at GECCO 2009 this summer. It studies how sensitive HyperNEAT is to different geometric...
Just for the record, I highly recommend this paper. It's the most extensive study published on the effect of varying geometry on HyperNEAT performance. Some...
Hi! Thanks for posting this paper, great work! By the way I also noticed that HyperNEAT performs well even in cases of a substrate that is messed up. My first...
We just put up a HyperNEAT Users Page: http://eplex.cs.ucf.edu/hyperNEATpage/HyperNEAT.html We have several reasons for making this page available, some of...
Peter, Jeff and I have had a lot of in-depth conversations about whether or not the good performance of randomized geometry has anything to do with its...
Hi Ken, I agree with pretty much everything you've said. In human vision much of the geometry part of the recognition problem is handled by succesive layers of...
Hello Ken and Peter- Please see my responses interspersed below. ... The position I took when discussing this with Ken was that 'the geometry of a problem'...
Hello Ken- I think this is a great idea. I also noticed a while back that there was not a definitive landing page on the Internet when someone searched for...
Another interesting implication of the ability to exploit seemingly scrambled geometry is that it may be a good omen for the future goal of evolving the...
Hi Jeff! It is an interesting and useful analysis of HyperNEAT's sensitivity to geometry in the quadruped locomotion domain. BTW, I will also be presenting a...
... Hi Vinod. ... Congratulations! The paper looks very cool. I wish you the best of luck. ... I looked at these videos a while back and found that most of the...
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.
