it is my second project with neuroph and neuronal nets. :d
Great to use this sort of java functions. :)
If someone is interested? It would be great to make something together or to share stuff :d
You see all information about there.
It is about to make neuronal nets visible. Why? Hm..... My way of thought is to show, if a neuronal calculation is the same like normal mathematical functions.
They are :D
And to make a nice menu to play and finally to print them on a thsirt :d
Hey this is very cool! :)
Send me some pics to put it on Neuroph site, and I'll link to your project too.
We'll make something together. Is it posible to visualize backprop?
By the way can send me the link for the code doing this?
shure, you can have the code :d
There is also a helper file, how to make it run :d
For now i am satisfied with this version. Let me know what you think.
On the facebook site
you can see screenshots.
You are welcome to be part of Floweuron ::lol::
If you like to play around attractors and stuff like this
I love it.
www.electricsheep.org and http://aphophysis.org
looks also nice, but i never tried it
http://chaoscope.org very cool!
maby i got you wrong. Do you like me to work with you together on Neuroph?
It would be great to do. There are a few reasons for me to do. I would have something to show companies. Thats the main reason. Also i am interested. Years ago i was studying artificial inteligenz....only one semester.
I downloaded neuroph2.7 and compared it to 2.6. Are you using different functionnames? I was not able to call some from 2.6?? Did i download the wrong .tar?
Yes it would be great to work together, and it doesent matter who works with who, it will be shared code for both of our projects. Its just I wont have enough time as I would like to dedicate to this, but I'm definetly very interested and excited to do this. I could provide you support for Neuroph and you can throw some fency visualizations for Next neuroph release. We canpublish them on Neuroph site too.
Where you want to start? Do you have any ideas? I have found an article about attractors and nets. For now i have some ideas on paper. They are about to show the outputs. You can move with the mouse to the attributes. Look at them in the second view, the attractor view. If the customer, :d, likes to train the net with these inputs, to lower, higher the output, the graph, view one shows also the whole changes of the net.
The methodes are still on paper. I can make a preview, for you and of course for me, with a small jsf programm and if it suits you, we can include it in the neuroph IDE.
sounds so great :)
Hi, sorry for delay in responding to this.
Its ok to start with outputs, to get general view of network function.
The idea with exploring is cool.
Next we can do the same thing with weights and errors.
And it will be incuded in Neuroph IDE for sure. We're working on the several vizualisation stuff at the moment and this will be great addon. Just make sure you're using Swing and Apache 2 compatible licence for all components so we can integrate.