Skip to content

Neural Network for recognition of handwritten digits.

Notifications You must be signed in to change notification settings

knutsa/Digit_Recognition

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

41 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Digit_Recognition

This is a C++ implementation of a neural network for digit recognition on images from the MNIST database. The highest achieved test accuracy is 95%. The code implements a simple Neural Network class that uses standard SGD with constant learning rate and mean square error loss, the class could be used for arbitrary classification problems but it has only been tested with the MNIST database.

How to run

  1. Got to utils.hpp and set the ROOT_FOLDER macro to the location of the code.
  2. Compile all files in the src folder and then run the executable, e.g
 g++ -O3 src/*.cpp -fopenmp -o main
 ./main

No libraries needed!

About

Neural Network for recognition of handwritten digits.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages