The workshop proposes an overview of robust optimization, a vibrant field in optimization which addresses uncertainty in optimization problems. Many practical optimization problems involve uncertainty, coming from measurement errors, errors in the variables, or uncertainty about future decisions. It is therefore important to be able to find, in a moderate computing time, solutions that trade-off optimality against some measure of risk involved in the optimal decision.
The robust optimization methodology seeks solutions that are robust (eg, guaranteed to remain feasible, or achieve some bound on the cost function) despite changes in the problem’s data. While most such problems are hard, the methodology now provides computationally efficient approximation schemes that offer a nice balance between performance and guaranteed quality of approximation. The field is relevant to many areas of applied science, including statistical estimation and inference, and control theory.
The workshop proposes a guided tour of the area, exploring various ways to describe uncertainty (from parameter bounds to random variables with partially known distributions), approximate the decision problem (with quality estimates), or addressing dynamic (control) problems where the uncertainty is partially revealed as time evolves. The workshop also includes several case studies in various application domains, ranging from signal processing, machine learning, communications, graph theory, circuit design, to finance and economics, logistics and operations research.
Aharon Ben-Tal, Chair
(Technion - Israel Institute of Technology)
Dimitris Bertsimas (Massachusetts Institute of Technology)
Jason Cong (University of California, Los Angeles (UCLA))
Laurent El Ghaoui (University of California, Berkeley (UC Berkeley))
Arkadi Nemirovski (Georgia Institute of Technology)