Skip to content

Test project for neural networks - Handwritten digit recognition on MNIST dataset

License

Notifications You must be signed in to change notification settings

ChriZ982/NeuralNetTest

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

32 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

NeuralNetTest

Main repository at: https://gitlab.com/ChriZ98/NeuralNetTest

This is my test project for the NeuralNetLib.

What is this repository for?

The task is to recognize handwritten digits. In this demo you can write some number on the screen and a neural network will tell you what it is. You can play around with different learning setups. My best network has an accuracy of 98% on the test data. Go ahead and try to beat my accuracy. ;)

What are neural networks?

Neural Networks are very small representations of the human brain. Every network consists of different layers and performs mathematical operations on them. The whole network can be seen as a black box acting like a function. Networks can be trained so that a specific input is resulting in a desired output.

Great literature and references

The following list contains links to great articles and other helpful content:

Version History

Version 1.0 - Neural Net

  • First version with working handwritten digit recognition.
  • First version of math lib.
  • Got working matrix and vector classes.
  • First version of neural network lib.
  • Implemented working stochastic gradient descent.

License

NeuralNetTest by ChriZ98 is licensed under a

License Image

Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License