PIMS- SFU Distinguished Visitor: Claire Boyer

The goal of these lectures is to present two problems arising in various applications when data is missing: compressed sensing and matrix completion. Both consist in reconstructing a very high-dimensional object from a few information. This is not possible unless we assume some structure prior on the object to reconstruct. They are closely related, that is why a theoretical overview given for the first one will be very useful for the study of the other. Furthermore, both can be relaxed as convex programs, which allows to design efficient resolution strategies. Theoretical key concepts for reconstruction guarantees will be detailed and an insight will be also given on the main algorithms used to tackle these problems.

Lecture Times:

Part I - Tue May 16 1:30-2:30

Part II - Wed May 17 1:30-2:30

Part III - Fri May 19 1:30-2:30