A0200
Title: Computing multiple-objective optimal regression designs via CVX
Authors: Julie Zhou - University of Victoria (Canada) [presenting]
Abstract: Model-based optimal regression designs with multiple objectives are common in practice. The objectives are often competitive, such as functions from A-, c-, D-, E-, and I-optimality criteria. It is extremely hard to derive analytical solutions for optimal designs with multiple objectives, and there are also no general and efficient algorithms for searching such designs for user-specified nonlinear models and criteria. We propose a new and effective approach for finding approximate multiple-objective optimal designs via the CVX solver. It can efficiently find different types of multiple-objective optimal designs after the optimization problems are carefully formulated as convex optimization problems. This approach is flexible and can be applied to any regression model. We present applications for minimax and efficiency constrained multiple-objective optimal designs.
