Generative Modeling

Our aim is to develop models and algorithms that can analyze and understand the treasure trove of data. Generative models are one of the most promising approaches towards this goal. To train a generative model we first collect a large amount of data in some domain (e.g., think millions of images, sentences, or sounds, etc.) and then train a model to generate similar data. The intuition behind this approach follows a famous quote from Richard Feynman: “What I cannot create, I do not understand.” —Richard Feynman

Staff

Fischer, Raphael
Saadallah, Amal