Neural Nettwork (ANN) Extension

[jrmrhrb00]

LitDev,

Now the program runs. At least on the screen.I don't get an .exe file in the bin folder though. I do get a .pdb file and 2 .dll files. All files that are created go in C:\Users\Administrator\source\repos\SmallBasicANN.
nothing gets put in C:\Users\Administrator\source\repos\TestANN. Shouldn't there be files in that? Also, any idea of why no .exe file?

JR

[litdev]

Mmm... Pretty sure it should be there? - it is somewhere. Anyway if it runs you can start to look at the code, debug through it and see how it works.

   

[jrmrhrb00]

LitDev,

I found it. Right after folder repos there was 2 folders, testann, smallbasicann. Testann had nothing nothing in it. Smallbasicann had the same testann folder inside of it and that is where the .exe was. I deleted the testann folder that had nothing in it.

I still need to figure a lot of this out!

Thanks for helping.

JR

[jrmrhrb00]

I am sure the program (MTTK615.000) isn't right, but I can't figure it out. The neural network should run what is there as an or network. So, for data it just contains an or data table. I would think that with the input values of the table that each one of the outputs from the network would be very close to 1, but only 2 of them are.

any help would be appreciated.

Thanks,

JR

[z-s]

On my pc it only shows Training Started.
and if I uncomment line number 11 it shows error :

index was outside the bounds of array.
at SmallBasicANN.ANN..ctor(String fn)
at SmallBasicANN.NeuralNetwork.Load(Primitive Filename)
at _SmallBasicProgram._Main()

don't know what is problem as I am weak in ANN

[jrmrhrb00]

Z-S,

I tested the program again and it runs. on mine it shows textwindow with training started and then the full display after about 30 seconds. I did this program as I am weak on ANN too! You don't need to uncomment line 11. Don't need it.

Thanks for helping,

JR

[AbsoluteBeginner]

(translated by Google translator)

Hello. 
If I look quickly, I saw three suspicious places.

First, in your code " NeuralNetwork.BinaryOutput(name,0,"False") " the parameter indicating that BINARY data is used is set to "False".
But, you are using BINARY input data.
Therefore, this parameter should be set to "True".

Secondly, you have commented out the line of code that should create the training data file: " 'File.WriteContents(trainingData,data) "
Uncomment it.

And thirdly, you are incorrectly creating training data and setting input data for the neural network to work.
Here is an example of usage: " NeuralNetwork.Use(Name, "0, 0") "

" input = "0, 0" "

[jrmrhrb00]

Z-S,

Thanks for help. I made the changes for the first 2 suggestions and it works correctly now.
 
How do I export the updated program? I see other people update with a "MTTK615.000.0", but if I just publish it again it just gives a new publish number. I figured it out. I needed to import it again and make the changes and then republish. So, the corrected program is "MTTK615.000.0".

I was thinking as binary as '00000000' not as 0 which is ascii, but it worked!

for some reason the import code didn't work, So, here's one that should "FPNQ191.000". Disregard the other one.

Thanks for the help,

JR

[jrmrhrb00]

I am trying to get the above program (FPNQ191.000) to work with binary data. The program above is using text for training and uses binary for the output. The below program was changed to using binary for the training and binary for the output. I don't see any reason why this does not work. The program with the changes is: (FPNQ191.000-2)

JR

[AbsoluteBeginner]

(translated by Google translator)

Hello,  jrmrhrb00. 

I see several mistakes in your code.

First, the structure of your neural network must have 4 outputs. The neuron at each of these outputs will be trained to recognize only one distinct combination of input values.
That is, the network structure should be "2, 4, 4".

Second, the training data should not be a single string of characters.
Each character of your training data must be the only one on the line.

And thirdly, to train the neural network in your program, ONE complete set of data is enough.
You don't have to repeat this set in a cycle 100 times.