Skip to content

Feed-forward neural network built from scratch trained to discriminate between hand-written digits. 28x28 pixel image samples from the MNIST dataset are used.

License

Notifications You must be signed in to change notification settings

mkeid/Neural-Network-For-Digit-Prediction

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Neural Network For Digit Prediction (Machine Learning)

Creates and trains a neural network to predict images (28x28 pixels) of hand-written digits.
The accuracy based on its own training examples is then computed and printed to the log.

Training example records in the mnist set consist of 29 elements separated by ","
Element 0 is the labeled class and the remaining 28 elements are pixel intensities for each image pixel.

The default structure of the neural network consists of:

  • 28 input nodes for each image pixel (calculated)
  • 1 hidden layer comprised of 30 nodes (hard-coded)
  • 10 classes for hand-written digits (calculated)

About

Feed-forward neural network built from scratch trained to discriminate between hand-written digits. 28x28 pixel image samples from the MNIST dataset are used.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages