An Overview of Metaheuristic Algorithms with a focus on Particle Swarm Optimization for Searching Efficient Experimental Designs
Time:15:00-16:00, December 28, 2022 / Place:M212
(講座教授,玉山學者)
The class of nature-inspired metaheuristic algorithms is increasingly used to tackle all kinds of optimization problems across disciplines. It also plays an important component in artificial intelligence and machine learning. Members in this class are general purpose optimization tools that virtually require no assumptions for them to be applicable. There are many such algorithms and to fix ideas, we review one of its exemplary members called particle swarm optimization (PSO).
We review optimal design theory and methods and discuss recent applications of PSO to find different types of efficient experimental designs. We provide resources, where codes for PSO and other metaheuristic algorithms and tutorials with examples are available. As applications, we use PSO to generate different types of dose response designs, early phase II clinical trial designs, and minimax or maximin optimal designs.