This talk will feature three techniques (linear and non-linear), namely: 1- PCA, 2- Deep learning based approach (Autoencoder), and 3- t-distributed stochastic neighbor embedding (t-SNE). Dr. Abdelmoula will show some applications on the common machine learning bench-marking dataset of MNIST as well as mass spectrometry imaging (MSI) dataset. He will show how t-SNE could help to reveal molecular intra-tumor heterogeneity as seen by MSI.