Skip to content

Repo for all neural nets stuff from my 10th grade neural nets class

Notifications You must be signed in to change notification settings

rishab-partha/Neural-Networks

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

32 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Plagiarism Notice

Do NOT copy any code from this repository. By using this code, you will be held liable for accusations of academic plagiarism.

Files

config.txt

Contains:

  1. Hyperparameter path
  2. Layer Configuration path
  3. Network Input/Output path
  4. Initial Weights path
  5. Final Weights path
  6. Path for a folder to write the outputs

inputs.txt

Contains number of inputs on first line. Then, every pair of lines has paths of inputs files on the first line and paths of output files on the second line.

layersizes.txt

Contains the number of layers on the first line. Then, every line after that contains the size of one layer.

NeuralNets.py

This is the original construction of the single-output trainable neural net for one hidden layer.

Command Line Prompts

--config_path - the file name of the configuration file

NeuralNets.pyproj

This gives the ability to configure and use the project on VSCode.

NeuralNetsMultipleOutput.py

This is the new construction of the multilayer perceptron with one hidden layer but multiple outputs

traininghyperparams.txt

Contains line by line:

  1. Max #iterations
  2. Initial Learning Factor
  3. Total error to terminate at
  4. Learning Factor Adjustment Rate (what to multiply and divide it by)
  5. Minimum total error decrease over a certain number of iterations
  6. Maximum learning factor
  7. Minimum lambda (essentially epsilon)
  8. Distance between iterations where learning factor and total error are printed and the total error decrease is checked
  9. A path to random range files
  10. The range of the inputs so that it can be scaled to between 0 and 1
  11. The range of the outputs so that the network can scale outputs between 0 and 1 to the expected output range

randomrange.txt

Each line contains a comma separated range of values for random values for the nth layer.

weights0.txt

With the definition of the weights as w[n][k][j], lists the input weights line by line sorted first by n, then k, then j.

weights1.txt

With the same definition as weights0.txt, weights1.txt represents the weights after training.

About

Repo for all neural nets stuff from my 10th grade neural nets class

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages