Academic

Handwritten Character Recognition

For this project, I compared feedforward and convolutional neural networks on handwritten English characters using Julia. I trained the models with the handwritten character dataset by Cohen, Afshar, Tapson, and van Schaik.

I looked at accuracy, confusion matrices, and learning curves to understand which characters each model recognized well and where it struggled.

A handwritten R alongside neural-network outputs; the predicted label is R.