Derives exponential upper bounds under the null for the spectrally truncated normalized MMD and supplies a practical data-adaptive quantile estimator with hyperparameter tuning that does not require splitting.
Kernel mean embedding of distributions: A review and beyond
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.ST 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Non-asymptotic two-sample kernel testing with the spectrally truncated normalized MMD
Derives exponential upper bounds under the null for the spectrally truncated normalized MMD and supplies a practical data-adaptive quantile estimator with hyperparameter tuning that does not require splitting.