更新时间：
报告人: 叶桂波 助理教授
University of Iowa
报告题目：Learning Gradients for variable selection and dimension reduction
报告时间：2013年06月13日下午15:30开始
报告地点：行政楼703
报告摘要: Variable selection and dimension reduction are two commonly adopted approaches for high-dimensional data analysis, but have traditionally been treated separately. In this talk, we first show that variable selection and dimension reduction are closely related by learning sparse Gradients. Firstly, we introduce a sparse gradient model which impose a sparsity constraint on the gradient. Variable selection is achieved by selecting variables corresponding to non-zero partial derivatives, and effective direction spaces are extracted based on the eigenvectors of the derived sparse empirical gradient covariance matrix. An efficient iterative soft-thresholding algorithm is developed to solve the sparse gradient learning model, making the framework practically scalable for medium or large data sets. The utility of SGL for variable selection and feature extraction is explicitly given and illustrated on artificial data as well as real-world examples. The main advantages of our method include variable selection for both linear and nonlinear predictions, effective dimension reduction with sparse loadings, and an efficient algorithm for large $p$, small $n$ problems.
报告人简介：叶桂波，2007年博士毕业于复旦大学，2007-2008年在香港城市大学及新加坡国立大学做访问学者，2009年至2012年在美国加州大学Irvine分校做博士后，2012年至今在 美国爱荷华大学(University of Iowa)数学系任助理教授。叶桂波博士的研究兴趣包括学习理论，凸优化及它们在生物信息学和计算生物学方面的应用，目前已发表SCI高水平论文10余篇。