A unified penalized method for sparse additive quantile models: an RKHS approach (Q2409400)

The paper puts forward an additive quantile regression model for high dimension and sparsity versus small sample size data with a novel sparsity-smoothness penalty under the reproducing kernel Hilbert space. The introduced model is optimized through a hybridization of the majorize minimization and the proximal gradient approaches. Additionally, oracle inequalities are set under weak conditions. The theoretical flow is accompanied by simulations and real-world test cases.