【报告摘要】This presentation covers two central topics: robust rank regression and transfer learning for quantile regression. The first part introduces a novel approach using a simultaneous feature- and sample-splitting ADMM algorithm for penalized rank and quantile regression, addressing the challenges of high-dimensional data. The method's theoretical foundations, including convergence properties, are presented, along with numerical experiments showcasing its practical effectiveness. The second part focuses on transfer learning for quantile regression, proposing a framework that transfers information across quantiles to improve estimation. This section explores the asymptotic properties of the proposed models and provides simulations that demonstrate their robustness in dealing with heteroscedasticity and outliers.