报告时间：2019年3月15日星期五下午2:30
报告地点：8A307
报告题目：A majorized proximal point dual Newton algorithm for nonconvex statistical optimization problems
摘要：In this talk, we consider high-dimensional nonconvex statistical optimization problems such as the square-root Lasso problem, where in the objective the loss terms are possibly nonsmooth and the regularization terms are difference of convex functions. We shall introduce a majorized proximal point dual Newton algorithm (mPPDNA) for these problems. Our key idea for making the proposed mPPDNA to be efficient is to develop a dual based sparse semismooth Newton method to solve the corresponding subproblems. By using the Kurdyka-Łojasiewicz property exhibited in the underlining problems, we prove that the mPPDNA algorithm converges to a d-stationarity point. We also analyze the oracle property of the initial problems used in our algorithm. Extensive numerical experiments are presented to demonstrate the high efficiency of the proposed algorithm.
报告人：孙德锋（Sun Defeng）教授（香港理工大学）
报告人简介：孙德锋教授现任香港理工大学应用数学系讲座教授。在此之前，孙教授担任新加坡国立大学数学系教授及风险管理研究所科研副所长。他主要研究连续优化，在矩阵优化及统计优化理论、算法及其应用方面取得一系列重要的突破性成果。孙教授完成了许多大规模复杂优化问题的软件，例如：通用的求解大规模半正定规划软件SDPNAL/SDPNAL+，协相关矩阵校准的程序，以及最新的适用于各种各样的统计回归模型的软件包SuiteLasso。由于他在优化计算方面的杰出贡献，他2018年获得了由国际优化学会每三年颁发一次的Beale—Orchard-Hays数学规划计算奖。目前孙教授研究集中于建立大数据优化和应用的下一代方法基础。孙教授还积极参加许多学术活动，2010-2013年担任Asian-Pacific J. Operational Research主编，2014-2017年担任Mathematical Programming Series B编委，现担任Mathematical Programming Series A，SIAM Journal on Optimization，中国科学-数学，Journal of the Operations Research Society of China，Journal of Computational Mathematics的编委。
