题目:Sparse Poisson Regression with Penalized Weighted Score Function
时间: 4月7日下午3:00-5:00
地点:明德主楼1030
摘要:I will talk on sparse regressions. Lasso and compressed sensing will be discussed. Especially, I will emphasize on our recently proposed new penalized method to solve sparse Poisson Regression problems. Being different from l1 penalized log-likelihood estimation, our new method can be viewed as a penalized weighted score function method. We show that under mild conditions, our estimator is l1 consistent and the tuning parameter can be pre-specified, which owns the same good property of the square-root Lasso. The simulations show that our proposed method is much more robust than traditional sparse Poisson models using l1 penalized log-likelihood method.
简介:
贾金柱,北京大学数学科学学院概率统计系和北京大学统计中心副研究员。2009年1月北京大学博士毕业。2009年1月至2010年12月,UC Berkeley博士后。
主要研究方向是高维统计推断和、统计机器学习和因果推断。在变量选择方法的理论研究、高维数据统计学习的应用和因果推断等领域发表论文多篇。主持国家自然科学基金青年项目和面上项目各一项。参加973重大项目一项,参加国家自然科学基金创新群体项目一项。