Peter, do you have a version of rtNEAT that compiles in Visual Studio 2008? It probably can't hurt to post that in the files system as well since I sometimes...
Thinking a bit more along the same path (pun intended) would be G code optimization. I don't know if you've ever watched a cnc mill wander around waisting time...
Hi ... This sounds similar to the travelling salesman problem - multiple points to visit and trying to minimize the overall distance travelled. So if the...
Hi all, I'm not sure if someone else has notice this "bug" in rtNEAT, but the max_depth() method in network.cpp has the potential to be forever in loop if a...
Hi Cesar, I know of this problem. It is obvious that depth cannot be determined in a recurrent network, but in general it depends, what is your way to handle...
I don't have VS2008 installed yet. I only guess that it should be still compatible, because it compiles with VS2005 without a problem. But I will try to...
... This sounds similar to the travelling salesman problem" Yes it does. The possible difference being that in some cases you might be working with 3D space....
Yes I have heard about this problem coming up in the XOR experiment. Most neuroevolution experiments are not classification experiments (i.e. they don't have a...
Ken, I am confused how you could end up in an endless activation loop. If you are moving from the input nodes forward through the hidden nodes to the output...
Hi Ken, While implementing NEAT-Python we had to solve the same problem and I wasn't sure how to do it efficiently for arbitrary feedforward networks (for...
Hi Peter! ... I agree if recurrence is allowed, which is not the case in problems like XOR or when I strictly need feedforward topologies. In these situations...
It is about the loop in which you calculate the longest path from an input node to an output node. Not the activation loop in which you activate the network. ...
I work with a number of manufacturing companies and often see a great deal of inefficiency in their processes. I have also struggled with PCB design programs...
Colin, The general size and restricted areas of the PCB would be manually determined, but the cppn should place most the parts in a way that lends efficient...
Peter, That doesn't make sense to me. Cesar's comments indicate he has a similar implementation to mine. Basically, every node's value in a single instance of...
It's not clear to me that it's desirable to calculate the number of times to activate by using the depth, regardless of how you handle loops. This method works...
Hey Jim, I think there is some confusion here because the issue in question is specific to the rtNEAT implementation (the bug may be present in the original...
Thomas, Were you addressing my response? Or are you referring to the original poster's endless loop implementation? My implementation has no notion of depth....
Jim, I think the problem resides in adding a link when applying the "depth strategy" for feedforward network activation (it's hard to see the right order of...
Joel, Ah. Okay. That's interesting. Well, I have not looked at the C++ code, so I don't know how it is implemented. I am confused what it means by activating...
... This is the same with NEAT-Python implementation, there's no concept of depth. The nodes are activated only once in the right order (the order is defined...
Jim, Computing the depth of a network does not have anything to do with the way you activate the network. By activating a network I mean just what you say, the...
This is an unusual kind of question for this group, but does anyone know anything about the recent activities or whereabouts of Mat Buckland? For those who...
I work primarily in Python so I was pleased to see two implementations on Google Code. NEAT-python (http://code.google.com/p/neat-python/) appears to be the ...
Hi shaflidason, The only reliable enough experiment on the repository is xor2.py. The double pole balancing needs some fixing, but I have a working version on...
Hi Cesar, Thanks for the reply. ... How about enabling the issue tracker on the project page and getting some of these tasks/bugs in there? I'm sure there are...
Cesar, Yes that does sound like a good approach- better than my approach for identifying recurrent connections in rtNEAT. My approach involves loop-checking,...
Hi Stefan! ... I thought it was enabled, sorry for that. The Issue tracker is available now. ... I'm not adding new functionalities to version 1.0, everything...
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.
