You are invited to attend** tutorial lecture**s by

**Prof. Yoram Bresler**

Coordinated Science Laboratory and the Department of ECE

University of Illinois at Urbana-Champaign

On the subject:

**Machine Learning for Inverse Problems in Imaging**

**Wednesday, June 12 2019, at 10:00-16:00**

Inverse problems, involving the recovery of an object from indirect measurements, are central in engineering and science. Most interesting inverse problems are ill-posed and their solution requires regularization. Classical approaches to the solution of such problems usually involve the solution of a variational problem of the fitting the reconstruction to the data using a model of the acquisition operator, under a penalty or constraints representing the properties of the underlying object or the desired solution, to regularize the solution. This description encompasses a broad range of inverse problems, including those that arise in compressive sensing, where the regularizer involves sparsity or low rank of some object-related quantities.

Challenges in this approach to solving inverse problems include (i) computational difficulty of solving the resulting optimization problems, which may be non-convex; (ii) difficulty to obtain good models of the object to be recovered to guide the design of a regularizer – especially if the object is natural, rather than a man-made; and (iii) in some applications, difficulty to fully characterize or describe analytically or computationally the measurement operator.

All these challenges are being recently addressed by a data-driven approach to inverse problems, in which the model for the data, the model for the measurement operator, and an efficient structure for computing the solution are all learned from the data. This progress is enabled by new problem formulation, theoretical results, and the recent advances in deep learning technology.

This tutorial will survey the state of the art in this field, introducing the key ideas, approaches, and results, illustrating on several applications in imaging.

**Room 011, Kitot Building, Faculty of Engineering**