README.md

character-recognition

A website demonstration of how linear algebra techniques (e.g. singular value decomposition) can be used to classify handwritten characters. An accompanying slideshow explains the mathematical intuition of the machine learning models developed for the project. The website is interactive, allowing users to draw a character and see the predictions of different models.

Commands

Install	Run	Clean
pip3 install -r requirements.txt	python3 app.py	rm -rf *.model
mkdir ~/.cache/emnist
wget https://biometrics.nist.gov/cs_links/EMNIST/gzip.zip
mv gzip.zip ~/.cache/emnist/emnist.zip

Credits

• Linear Algebra with Mr. Honner

• Naoki Saito (MAT 167 @ UC Davis) - Lecture 21: Classification of Handwritten Digits

• MachineLearningMastery - CNN for MNIST

• 3Blue1Brown (YouTube) - Neural Networks

• Samson Zhang (YouTube) - Neural Network from Scratch

• Azka Redhia (Medium) - Neural Network for Handwritten Digit Recognition

• Kaggle - EMNIST (Extended MNIST)

• Achintya Tripathi (Kaggle) - 97.9% accurate EMNIST code

• Cohen, G., Afshar, S., Tapson, J., & van Schaik, A. (2017). EMNIST: an extension of MNIST to handwritten letters. Retrieved from http://arxiv.org/abs/1702.05373