The Division of Science and Technology (DST) held a lecture on “A Polynomial Algorithm for Best Subset Selection” given by guest speaker Prof. Rongxian Yue on November 12th, 2021. Prof. Yue introduced G-optimality condition for designing mixed-effects models and proposed the competitive swarm optimizer (CSO) algorithm to generate G-optimal designs for linear mixed models.

Prof. Yue is giving the lecture

Prof. Yue is the professor and doctoral supervisor of Shanghai Normal University and currently serves as the vice president of the Experimental Design, and Survival Analysis Branches of the Chinese Association for Applied Statistics. He is the leader of 5 General Programs of the National Natural Science Foundation of China.

Prof. Yue established an equivalence theorem for confirming G-optimality of an approximate design, which simplifies the original multiple nested optimization problems. Prof. Yue pointed out the difficulty of finding and studying G-optimal designs analytically and proposed a meta-heuristic algorithm called competitive swarm optimizer (CSO).

Prof. Yue and his team applied the CSO algorithm to generate G-optimal designs for linear and generalized linear mixed models. The results indicated that the approximate designs met the G-optimal condition well.

Prof. Yue answering audience’s questions

At the end of the lecture, Prof. Yue interacted with the audience and answered their questions.

Prof. Kaitai Fang (left) presenting UIC souvenir

Prof. Yue and UIC teachers

Reporter: Yawei Peng

Photographer: Guoqiu Zhang, Yawei Peng

Editors: Ms. Garbo Hu & Ms. Guoqiu Zhang