A0647
Title: Numerical algorithm-aided approaches for analytically finding optimal designs
Authors: Ping-Yang Chen - National Taipei University (Taiwan) [presenting]
Abstract: The traditional way in statistics to find optimal designs for regression models is an analytical approach. However, the mathematical technique often faces challenges with complex experimental setups or difficult-to-solve criteria, suggesting that developing flexible and effective algorithms to search for optimal designs is highly valuable. In particular, numerical results from an algorithm can be instrumental in deriving analytical descriptions of optimal designs. For instance, particle swarm optimization has been shown to be quite effective for finding optimal designs for hard design problems. How the output can be utilized is demonstrated to discover new analytic optimal designs, showcasing its utility in scenarios such as A-optimal designs for generalized linear models and the standardized maximin D-optimal designs for nonlinear inhibition models. These examples illustrate how optimization algorithms complement traditional analytical methods, enhancing the identification of optimal experimental designs.
