This is a configurable Neural Network written in C#. The Network functionality is completely decoupled from the UI and can be ported to any project. You can also export and import fully trained networks and datasets.
Great job! I couldn't find any method that helps to write/read the trained network to/from the file.
That would be a great option to have because normally training might take few hours and you might not want to do that everytime you start over.
I left your network file, NetworkExample.txt, alone.
I modified your DatasetExample.txt file such that Targets = V1 + V2.
I tried to Train to Minimum Error = 0.1..
It converges to error = 0.7501.
It never trains to 0.1.
What is wrong?
How do I fix this?
Friendly hello.
I think your project is a good start to who don't know all the thing in Neural Networks but want to know more and practice , we all appreciate that.
I began to check out your code and notice that you are using the new C# 6.0 syntax, that syntax just work for the VS2015 compiler.
So I think that it will be good to tell that in the REDME file, so people dont have problems when compiling.
In the Program.cs there is a SetupFromFile() method, which does the splitting of a fileContent. When it splits the fileContent with Environment.NewLine first parameter, the new line splitter is \r\n in my environment and so the function doesn't split the fileContent. It causes expected error with unability to Load all data from file.