学术报告第201854期

报告题目：Convexity and Regularization Issues in System Identification

报告人：Lennart Ljung（院士）

报告时间：2018年9月10日 下午14:30—15:30

报告地点：南一楼中311

报告摘要：

System Identification is about estimating models of dynamical systems from measured input-output data. Its traditional foundation is basic statistical techniques, such as maximum likelihood estimation. This relies on minimization of criterion functions that typically are non-convex, and may cause numerical search problems. Recent interest in identification algorithms has focused on techniques that are centered around convex formulations The development concerns issues of regularization for sparsity and for better tuned bias/variance trade-offs. A quite different route to convexity is to use algebraic techniques to manipulate the model parameterizations. Both these aspects are illustrated in the presentation.

报告人简介：

Lennart Ljung was born in 1946 in Malmö, Sweden. He attended Lund University and earned a B.A. in Russian Language and Mathematics in 1967, a M.S. in Engineering Physics in 1970, and a Ph.D. in Automatic Control in 1974. Ljung served on the academic staff at the Lund Institute of Technology from 1968 until 1976 when he became Professor of Automatic Control at Linköping University in Sweden. He served as the head of the Control division and Chairman of the Department of Electrical Engineering at Linköping Institute of Technology and Director of the NUTEK/VINNOVA Competence Center ISIS, and is currently Director of the Strategic Research Center MOVIII. Ljung’s research focuses on model building, system identification and adaptation, resulting in numerous degrees and awards, several publications and articles, and extensive contributions to the field of Control Theory.