5 problems

Uncertainty-quantification conjecture. Such a weak uncertainty-quantification condition would not generally hold without the parametric modeling condition ${\rm (A3)}$.

The paper uses a piecewise-polynomial presmoothing estimator in its two-step procedure. Least-squares splines are an alternative smoothing method that would avoid discontinuities a…

The paper considers sparse high-dimensional additive models with component functions $f_1,dots,f_q$, covariate density $p$, and smoothness parameter $\rho_1$. Its minimax theorem e…

Consider backfitting estimation for general additive models, using kernel-based estimation methods such as Nadaraya–Watson or local polynomial smoothing. Backfitting convergence co…

Consider the additive model … with $E(\varepsilon\mid U,V)=0$ almost surely, and estimate its component functions by backfitting, using kernel smoothing. OR's convergence conjectur…