嘉宾简介：2014年于南开大学获得博士学位。2015年于美国佛罗里达大学做博士后。曾访问香港浸会大学、新加坡国立大学、香港大学。主要研究方向包括统计过程控制、非参数模型、高维数据分析等。曾在著名统计期刊AOS,JASA,Biometrika，Statistica Sinica上发表多篇高水平论文。

摘要：The generalized likelihood ratio (GLR) statistic (Fan, Zhang, and Zhang (2001)) offered a generally applicable method for testing nonparametric hypotheses about nonparametric functions, but its efficiency is adversely affected by outlying observations and heavy-tailed distributions. Here a robust testing procedure is developed under the framework of the GLR by incorporating a Wilcoxon-type artificial likelihood function, and adopting the associated local smoothers. Under some useful hypotheses, the proposed test statistic is asymptotically normal and free of nuisance parameters and covariate designs. Its asymptotic relative efficiency with respect to the least squares-based GLR method is closely related to that of the signed-rank Wilcoxon test in comparison with the t-test. Simulation results are consistent with the asymptotic analysis.