Handwritten Digit Recognition using Edit Distance-Based KNN
Marc Bernard, Élisa Fromont, Amaury Habard, Marc Sebban · 2012
We discuss the project given for the last 5 years to the 1st year Master students who follow the Machine Learning lecture (60h) at the University Jean Monnet in Saint Eti-enne, France. The goal of this project is to develop a GUI that can recognize digits and/or letters drawn manually. The system is based on a string representation of the dig-its using Freeman codes and on the use of an edit-distance-based K-Nearest Neighbors classifier. In addition to the machine learning knowledge about the KNN classifier (and its optimizations to make it efficient) and about the edit distance, some programming skills on how to develop such a GUI are required.