Establishes CMMD as a family of kernel-based metrics for differences between conditional distributions, with levels 0-2 and general s, plus a doubly robust estimator consistent if at least one model is correct.
Advances in Neural Information Processing Systems , year =
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
-
Measuring Differences between Conditional Distributions using Kernel Embeddings
Establishes CMMD as a family of kernel-based metrics for differences between conditional distributions, with levels 0-2 and general s, plus a doubly robust estimator consistent if at least one model is correct.