LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

Neural network in LabVIEW - I'd like to share the code

Hi Everyone,
 
I have developed a fully configurable multi-layer perceptron (artificial neural network), feed-forward, back propagated. I used LabVIEW 5.1 and wrote the whole thing from scratch. The program (aNETka ver. 1.0) incorporates: biases, autoscaling, automatic learn rate, momentum, etc.
 
You can download the zipped file from: http://www.geocities.com/zureks//aNETka.html
 
UPDATE: The website will soon go off-line because Yahoo shuts down the geocities. You can email me directly then: zureks@gmail.com
 
There is a operating manual, too Smiley Happy
 
Please post some comments if you use it...
 
Enjoy
 
Yodo
---------------------------------------------
Everything should be made as simple
as possible - but not simpler.
Albert Einstein
Message 1 of 25
(12,783 Views)
Hi Yodo
 
Basically it looks quiet good. The only thing I really get annoyed of is the huge diagram. Did you use a beamer with resolution 16000 x 12000? It's quiet hard to see how it works.
 
Thomas
Using LV8.0
--------------------------------------------------------------------
Don't be afraid to rate a good answer... 😉
--------------------------------------------------------------------
Message 2 of 25
(12,763 Views)
oops, guess there was a lot of interest in your program, you have exceeded you transfer allocation! And on a "non-Naughty" site!
Putnam
Certified LabVIEW Developer

Senior Test Engineer North Shore Technology, Inc.
Currently using LV 2012-LabVIEW 2018, RT8.5


LabVIEW Champion



Message 3 of 25
(12,760 Views)
The resolution I used was 1024x768. I know the diagram is huge, but I found that using subVI's slows down the algorithm a lot (in fact, the first version was one screen diagram with neat subVI icons). I took me a lot of time to rebuild it into non-subVI version, and since I did I just kept this mode. The most important bit is the main While loop - i tried to put the smaller pieces into Sequence or Case structures to be able to see what's going on. If you really want to see how it works just make a subVI's from the structures I provided, and then you can scale the whole diagram down.

I also submitted this example code to NI, but I can't see it (search engine doesnt find it). Just be patient with the free website I provided. Should work every 1h or so Smiley Wink


Yodo
---------------------------------------------
Everything should be made as simple
as possible - but not simpler.
Albert Einstein
0 Kudos
Message 4 of 25
(12,753 Views)
Thanks for sharing your work. Again something I tried to develop some years ago, but I never passed the original Perceptron level (single layer network). That thing was already smart enough to recognize numbers. 🙂
I'll try to run your vi on a few data (that could be the E-List...) as soon as possible.



That's worth 5 stars !..
Chilly Charly    (aka CC)
0 Kudos
Message 5 of 25
(12,734 Views)
Thanks for the words of appreciation. It was really hard thing to do. I got lost with the first version and I had to start completely from scratch again, because there was no way I could debug it.

I am sorry for not very "clean" coding, but for me it was enough and too much to get it working...

Regards

Yodo
---------------------------------------------
Everything should be made as simple
as possible - but not simpler.
Albert Einstein
0 Kudos
Message 6 of 25
(12,728 Views)

Thanks!

I will look at the code later.

JLV

Message 7 of 25
(12,720 Views)

Dear all,

I have visited the pointed URL, however no download option was available. So, I sent a message to the indicated e-mail, but I have not receive an answer yet (it is about a week now). Do you know if there is an alternative URL or  e-mail adrress, or any other communication way to ask for a copy of the toolkit. Do you know any other LabVIEW implemented neural network training tools that someone could try or even buy.

 

Thanks!!!

Message 8 of 25
(12,628 Views)
The website will soon go off-line because Yahoo shuts down the geocities. You can email me directly then: zureks@gmail.com
0 Kudos
Message 9 of 25
(10,874 Views)
NI has a community section now which makes it easier to post documents and files. You could place it there.

___________________
Try to take over the world!
0 Kudos
Message 10 of 25
(10,848 Views)