Derives adaptive Sobolev-norm learning rates for conditional mean embeddings in misspecified RKHS settings, achieving uniform convergence in some regimes.
Thus, we have, that there exists a K > 0 not depending on n or δ, such that: ||ˆCY |X − CY |X ||γ ≤ K log(δ−1)λ β −γ 2 n with probability 1 − 2δ
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ML 1years
2021 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Sobolev Norm Learning Rates for Conditional Mean Embeddings
Derives adaptive Sobolev-norm learning rates for conditional mean embeddings in misspecified RKHS settings, achieving uniform convergence in some regimes.