Speaker: Yifan Wang, PhD Candidate, Department of Statistics, Iowa State University

Title: Bimodel kernel and high dimensional inference under weak sparsity 

Abstract: The high dimensional statistics receive a lot of attention. There are a lot of aspects that could bring our attention. The theoretical aspect of high dimensionality is also worth mentioning. The local polynomial could help us focus on local solutions to help explain the variation which comes from local. But it fails when we meet correlated errors. This work gives some extensions from one- dimensional to high-dimensional. The strong sparsity has a lot of finished theoretical properties. This work applies weak sparsity to the disparity lasso estimator to reach some theoretical properties. We will also show our theoretical methodology could help to construct the confidence interval. We also use this disparity lasso estimator to form some tests in which we could set some threshold on statistics to improve the power.