Abstract

Nature-inspired meta-heuristic algorithms are increasingly studied and used in computer science and engineering research to solve high-dimensional complex optimization problems in the real world. It appears relatively few of these algorithms are used in main stream statistics even though they are simple to implement, very flexible and frequently able to find an optimal or a nearly optimal solution quickly. These general purpose optimization methods usually do not require any assumption on the optimization problem and the user only needs to input a few easy-to-use tuning parameters.

In this talk, I describe and demonstrate the usefulness of one of these algorithms called particle swarm optimization for finding different types of optimal designs for nonlinear models.  Several biomedical applications will be discussed, including one for constructing a multiple-objective optimal design for a dose response study. 

Key words: approximate design, exact design, equivalence theorem, information matrix, multiple-objective optimal design.