Introduces Δ-VFE pivoted Cholesky, a pivot rule maximizing the one-step gain in a VFE functional for kernel matrices via closed-form decomposition and batch sampling, yielding improved GP objective values and accuracy at low ranks.
cc/paper_files/paper/2007/file/ 013a006f03dbc5392effeb8f18fda755-Paper
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
Conditional KRR reduces to KRR on a residual kernel with an added O(1/sqrt(N)) term in expected test risk and outperforms standard KRR when the F-component is dominant.
citing papers explorer
-
Variational Free Energy Pivot Selection for Pivoted Cholesky
Introduces Δ-VFE pivoted Cholesky, a pivot rule maximizing the one-step gain in a VFE functional for kernel matrices via closed-form decomposition and batch sampling, yielding improved GP objective values and accuracy at low ranks.
-
Conditional KRR: Injecting Unpenalized Features into Kernel Methods with Applications to Kernel Thresholding
Conditional KRR reduces to KRR on a residual kernel with an added O(1/sqrt(N)) term in expected test risk and outperforms standard KRR when the F-component is dominant.