NNGP and GPnn achieve universal consistency, attain Stone's minimax L2-risk rate n^{-2α/(2p+d)}, and exhibit asymptotic robustness to hyperparameter choice under mild regularity assumptions.
Letf(u) =u ρ 2(u) andg(u) =L 2 u2p+1, thusf(0) =g(0) = 0
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ML 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
The Theory and Practice of Highly Scalable Gaussian Process Regression with Nearest Neighbours
NNGP and GPnn achieve universal consistency, attain Stone's minimax L2-risk rate n^{-2α/(2p+d)}, and exhibit asymptotic robustness to hyperparameter choice under mild regularity assumptions.