New Algorithms in Information Science

Stanley Osher
Institute for Pure and Applied Mathematics

Bregman distance has been around since 1967. It turns out to lead to unreasonably effective and somewhat familiar optimization techniques involving L1, BV and related penalty terms which arise in a host of new and important algorithms, e.g., compressive sending, nonlocal means. I'll discuss this and give several application, including hypperspectral and medical imaging.

Back to Workshop V: Applications of Optimization in Science and Engineering