Develops convMMD, a noise-convolved variant of MMD that preserves metric properties and enables consistent inference under known heteroscedastic measurement error.
For XDGMM, we provide the true error standard deviation in case of homoscedastic noise processes as a parameter to the method
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ME 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Convolutional Maximum Mean Discrepancy for Inference in Noisy Data
Develops convMMD, a noise-convolved variant of MMD that preserves metric properties and enables consistent inference under known heteroscedastic measurement error.