In order to begin ANN training, test data is needed. Given that this type of classification method is supposed to be good at classifying distorted or noisy images, we decided to attempt to classify the images here:
They are hand-crafted text files that contain shaped blocks, created from the characters 1 and 0. When they are stored on HDFS, the carriage return characters are removed so that the image is presented as a single line vector. So, the ANN will be classifying a series of shape images and then will be tested against the same images with noise added to determine whether the classification will still work. There are ...
