A sample implementation of a neural net classifier that recognizes hand-drawn digits.
Detailed documentation on the mathematics used here can be found in nnclassifier.ipynb
Clone and install nnc
from source
git clone https://github.com/joeyshi12/neural-net-classifier.git
cd neural-net-classifier
pip install -e .
Test the accuracy of one of the provided trained models
nnc view models/example1.model
Controls
- Clear canvas: c
- Guess digit with model: <space>
Train your own neural net model using the provided mnist data set
nnc train data/mnist.pkl.gz myclassifier.model
View other options with nnc -h