mnist-digit-recognition-feedforward

Implementing handwritten digit recognition on the MNIST dataset using a feedfoward neural network.

Getting started

Tested on Python 3.7.

Installation

1. Create a virtual environment venv in the current working directory using python3 -m venv venv.
2. Activate the virtual environment through source venv/bin/activate.
3. Install the dependencies from requirements.txt using pip install -r requirements.txt.

Usage

python3 predict.py -e 30 -l 32 16 -t 20

Sets up a neural network with hidden layers of [32, 16]. Trains using 30 epochs. Runs through 20 validation cases after training.

Acknowledgements

The eBook on Neural Networks and Deep Learning helped guide me through building and understanding the concepts of how a neural network functions.