Robust Optimization in Radiation Therapy

This lecture covers the concept of intensity-modulated radiation therapy, focusing on robust optimization problems in the context of uncertain breathing states. The instructor explains how to minimize dose variations in tumor voxels by optimizing beamlet weights. The lecture also discusses the trade-off between nominal and robust solutions, highlighting the impact on tumor underdosing and organ dose reduction. Additionally, the lecture delves into robust support vector machines, providing insights into worst-case hinge loss bounds and the role of regularization. The session concludes with a discussion on dynamic robust optimization, linear decision rules, and their applications in production planning.