Hey, Why it is that when NEAT (or maybe just sharpNEAT, what I'm looking at) runs networks, the input neurons' signals don't get changed even though there are...
I - PREMISE: NEAT in it's present state is metaphorically the wheel. In it's fundamental form, it was useful for various tasks. But the greater utility was...
... That's a general feature of NEAT, at least most implementations. The input neurons are different from other neurons in that their only role is to emit ...
... There shouldn't be any connections going *to* input nodes, only from them. If you are seeing such connections then that is a bug. The input neurons in...
Hi Oliver There isn't a UNIX based run script that comes with neat4j, it's on my todo list. The best thing to do is convert the batch files into shell ...
Hi Oliver, I've converted the two batch files - xor_train.bat and xor_test.bat for my Slack box. They are running fairly okay. I guess they should be okay for...
Hello there. I am using a modified version of the rtNEAT code release, to evolve roving eyes that identify grayscale images. It is very similar to the active...
Since this is the second question that's really more about C# programming than NEAT, please tell me if it's inappropriate to ask on this forum... If not,...
Yeah, I found it interesting to evolve a NN that would be better at learning, so I did some experiments on it with BackProp and NN learning to graph...
Hi I'm fairly new to NEAT. I'm thinking of writing my own NEAT implementation from scratch, rather than modify one of the existing releases. The main reasons...
Here's a few pointers: -Debug every function as soon as you create it. For exampe, when you write mutate_add_node, apply it to some dummy genomes and make ...
Ken, Thanks for the good advice. My idea to implement a delay in the firing of a node is simply to allow NEAT to insert a special node type with a single input...
Yeah, I read about Cascade Correlation for the first time in a class this semester. It seems like a good early step in algorithms for exploring good ...
There are variations with recurrence and other things, they're on the first link I've posted. But at a first glance, I agree that NEAT is more powerfull (by...
Hi Mike, yes I think that's reasonable. I haven't seen such a node attempted before, but it will be interesting to see if it serves a useful purpose. ken ... ...
I think a significant difference is that CC is geared more naturally to supervised training problems. That is, examples must be presented to the CC network as...
... Or to take it to the logical (?) extreme: write a piece of test code that will show the code to be working before you actually write the code. That way...
... this ... exploring good ... something ... stagnates you ... problem or ... There was something of a religious conflict about a decade back between the...
Ken, Two related questions: a) Would it make sense to implement last_activation in NNode as a vector and utilize a trait to evolve how many steps a delay is ...
Hello all- I am seeking a recommendation for the equivalent of a neural networks for dummies book. I have been doing grad research in evolutionary computation ...
Hi. I want know more about JNEAT, actually I am working in a robot projec, and I found JNEAT an interesting option, but I dont know how to use it. So if you...
Jeff, I highly recommend Blondie24 by David Fogel. It is an introduction to neural networks, GAs, and using the two together. It's also a good introduction...
Thanks for the suggestion. Does it take you through enough of the details of the various learning mechanisms (e.g. backprop) to really understand how they work...
PS. If anyone knows of a good web tutorial instead I could also print that out and take it with me. Thanks. Cheers, Jeff Clune Digital Evolution Lab ...
Hi Jeff, for the past few months I've been catching up on the NN topic. I found a great book by Hagan, Demuth and Beale, "Neural Network Design". They go...
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.
