Using a multi-class Logistic Regression and a Neural Network with regularization to identify handwritten digits
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
README.md
displayData.m
fmincg.m
handwritten.mat
lrCostFunction.m
nnCostFunction.m
nnPredict.m
oneVsAll.m
predictOneVsAll.m
randInitializeWeights.m
run.m
sigmoid.m
sigmoidGradient.m

README.md

machine_learning_digit_recognition

Using a multi-class Logistic Regression and a Neural Network with regularization to identify handwritten digits

Developed and written by Arnold Yeung

This project runs multi-class logistic regression and neural network with regularization for a dataset containing the pixels of handwritten digits. The dataset used is a subset of MNIST handwritten digits (http://yann.lecun.com/exdb/mnist).

This project is based on Exercises 3 and 4 in Coursera course, Machine Learning by Andrew Ng, Stanford University (https://www.coursera.org/learn/machine-learning).

All scripts and functions attached in this project, with the following exceptions, were written by Arnold Yeung:

  • fmincg.m
  • displayData.m

The main pipeline script is run.m

For more information, please visit www.arnoldyeung.com

If you have any questions or comments, feel free to contact me at contact@arnoldyeung.com