Seminars & Events

CSE Community Seminar | September 13, 2024

Presenter Laurent Demanet, Professor of Applied Mathematics, Department of Mathematics at MIT

Talk Title Recovery phenomena with symmetric autoencoders

Abstract

Can we find the solution of a PDE we have never seen, if we collect enough solutions of nearby equations? Is it possible to guess what a scene in an image would look like if the picture was taken from a different angle? Would it sound like you if an AI generated a deepfake of your voice?  These questions fit in a common mathematical framework of estimation of low-dimensional latent processes under maps of controlled complexity. I will explain how ML has advanced to answer these questions, possibly with mathematical guarantees in the form of recovery principles à la compressed sensing. Joint work with Pawan Bharadwaj, Matt Li, and Borjan Geshkovski.