A project related to the course named Machine Learning at The University of Trento. With a ML classification algorithm optical character recognition (ocr) is performed and evaluated. Dataset consist of handwritten characters from 16x8 bitmap images:
Training set instances: 41721
Test set instances: 10431
With a fixed C at 10, and the best gamma of [1e-1, 1e-2, 1e-3, 1e-4] found by grid search, the accuracy is 0.8996261144664941.