This was a convoluted neural network created for the Digit Recognizer competition on Kaggle. In the competition it scored a perfect score of 1.00000, placing me in the top 2% of participants. It uses both the provided competition dataset and the MNIST dataset for training.
These instructions will get you a copy of the project running on your local machine for development and testing purposes. See below for the prequisites before starting the installation.
- Python 3.x
- Keras (pip install keras)
- Numpy (pip install numpy)
- Pandas (pip install pandas)
- Download and unpack the zip file of this repository.
- Edit the file where necessary, such as reducing the epochs for testing purposes
- Open terminal and cd into the folder where you saved it.
- Run main.py in terminal.
$ python main.py
- Wait for completion of the training and the resulting csv to be produced.
- Fhoughton - Initial project
See also the list of contributors who participated in this project.
This project has no license and is free to be used and edited by anyone within the terms of the licenses provided with the libraries used.
- Thanks to the Kaggle community for the great help provided
- A great thanks to the developers of Keras, Numpy and Pandas for making this possible