Robust Optimization: Radiation Therapy & Support Vector Machines

This lecture covers robust optimization applied to intensity-modulated radiation therapy, where the goal is to minimize the expected dose received by all voxels while accounting for uncertainty. It also delves into robust support vector machines, showing the connection between robustification and regularization. The speaker explains how the worst-case hinge loss is bounded by the sum of the nominal hinge loss and the regularization term. Additionally, the lecture discusses robust convex optimization problems and dynamic robust optimization, highlighting the use of linear decision rules to simplify complex problems.