We analyze the handwritten AND image dataset from various writers using Variational Autoencoders, Siamese Network and then pass it through a Multi-Task Learning process to get the 15 features of the images directly. This is then used to compare the handwriting of the writers