Skip to content

This is a simple Neural Network model, that is written in almost raw python, without additional "precooked" libraries.

Notifications You must be signed in to change notification settings

Sachar04/Simple-NN

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 

Repository files navigation

Simple clear Neural Network without predone libs

Thoughts

This, is my second project, where I tried to make a ML model or Neural Network. This one was completly from scratch, and is currently my most succesfull and accurate ML-Model (if compared to ML-Project from earlier)

The Model

The current model, is a simple one and uses Sigmoid function, or (Logistic Regression) to train on some data {in my case repeted itterations of logical operations in binary-format e.g:[0 or 1])

It is therefore very powerfull, thus the data is simple, can be iterated more time if you need better precission.

If you need more operations, provide a testing data set in array-like form like:

here an example of adding AND opearations

dataset_and = [[[0,0], 0], [[0,1], 0],[[1,0],0],[[1,1],1]] //add a dataset
test_data = [[0,0], [0,1], [1,0], [1,1]] //add the "results" dataset that provides 1. input, 2. expected output
new_nn_and.train(dataset_and, 100000) //configure the amount of cycles you want the NN to learn (10k are mostly enough, from there the precission will not increase drastically)
new_nn_and.test(test_data, 'AND') //test it and print when done

TODO's

I have actually some big plans on this model, and i will try to add some color-related operations for fun and skill improvement and i will then update the project.

Remarks

  • If you have some suggestions, or find errors, feel free to add some pull requests, and i will review them ASAP :)
  • The project is completely open-source and u can use it however you like

About

This is a simple Neural Network model, that is written in almost raw python, without additional "precooked" libraries.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published