【报告摘要】Nonconvex optimization problems are ubiquitous in system identification, posing significant challenges due to the presence of multiple local minima. In this talk, we present several important identification tasks—ranging from linear to nonlinear systems—where, somewhat surprisingly, the underlying nonconvex formulations admit efficient global optimization. We reveal structural properties and algorithmic mechanisms that enable the computation of global minima with provable guarantees.

 【报告人简介】牟必强，中国科学院数学与系统科学研究院副研究员，于2008年从四川大学获得工学学士学位，于2013年从中国科学院数学与系统科学研究院获得理学博士学位，研究兴趣包括系统辨识、机器学习、目标定位跟踪。