Handwritten Digit Recognition

Background

Optical character recognition has been researched and used since the early 1900s to convert printed text into machine processable data, see Optical Character Recognition (OCR). The ability to process fixed font characters is well established but accuracy is hampered by the amount of noise in the image. Hand written, hand printed actually, characters are a much more difficult problem than fixed font due to the noise and lack of consistency.

Project Description

Our project is to train an autoencoder neural network to recognize hand written digits from the NMIST database with an accuracy greater than that achieved in a previous work, "Use Neural Networks to analyze handwriting Format", by Julia Nomee, Avery Leider, Stephanie Haughton, and Yahia Saeed, May, 2016. Our goal is to exceed the accuracy obtained in that work of 96.4%.