How much is an idea really worth? What defines success for a product? How can we quantify "better" or “worse”? At Microsoft we have tens of thousands of engineers and data scientists trying to improve products that touch over a billion people worldwide. The data scale is enormous, and we're trying to learn from that…

Convex optimization problems appear naturally across the sciences in fields such as compressed sensing, statistics, machine learning, image processing, and inverse problems. This talk will discuss the theory, applications, and methods of proximal minimization: a wide-reaching subset of convex optimization. Advantages: you do not need your optimization problem to be differentiable (e.g. minimization problems with l-1…