学术讲座-Trace Lasso Regularization for Adaptive Sparse Canonical Correlation Analysis with Applications

讲座题目

Trace Lasso Regularization for Adaptive Sparse     Canonical Correlation Analysis with Applications

讲 座 人

彭拯

讲座人

职称、职务

教授

主持人

许弘雷

讲座类型

R自然科学

讲座对象

全校师生

举办时间

2018/1/12 13:00

□社会科学

举办地点

福建省汽车电子与电驱动技术重点实验室会议室

讲

座

人

简

介

彭拯，现为福州大学数学与计算机科学学院教授，博士生导师，当前主持国家自然科学基金一项。

June     2008:  Ph. D., Department of Mathematics, Shanghai University
      June 2003: M.Sc., Department of Mathematics, Hunan Normal University
      June 1991: B.Sc., Department of Mathematics,  YueYang  Normal College
      June. 1998: B.Sc., Department of Computer     Science,  Xiangtan  University

讲座

主要内容

By adapting the trace Lasso and Lasso     regularization, an adaptive sparse version of CCA (adaptive SCCA for short)     is proposed. The adaptive SCCA reduces the instability of the estimator when     covariates are highly correlated, and thus improves their     interpretation.  The adaptive SCCA     model is reformulated to an orthogonality constrained optimization problem,     and an effective splitting method is proposed for solving the resulting     problem, The performance of the proposed SCCA model is compared with other     sparse CCA techniques in different simulation settings, and the validity is     also illustrated on the real genomic data sets.