search

CharlesHEaster / NeuralNet

My attempt to create a framework to create and train neural networks through fitness evolution.
0 stars 0 forks source link

Trial with history #7

Closed CharlesHEaster closed 3 years ago

CharlesHEaster commented 4 years ago

Possibly network and inputNode too. trial.histStrut[ ]. importHistInputs() makeInputs()

CharlesHEaster commented 4 years ago

Implimented on the input nodes. Now just to take it up through the levels to Trial

CharlesHEaster commented 3 years ago

Should be done now. Also, Trial Inputs are now an ArrayList<ArrayList> (from Double[ ][ ]). Refactoring that was annoying and I can't even remember now why I wanted to do it.

CharlesHEaster commented 3 years ago

ended up needing to change all double to Double. That was frustrating. Also, when I was finding the output from the min and max there was a scenario where I was trying to divide 0 by 0. extra fun. Kept getting NaN in a Double. Then I started using the built in debugger. good to learn