Under a low-dimensional subspace assumption, rescaling informative slices of the sliced-Wasserstein distance reduces to one global constant, so the classical SWD with a tuned learning rate is competitive with specialized variants.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Understanding Learning with Sliced-Wasserstein Requires Rethinking Informative Slices
Under a low-dimensional subspace assumption, rescaling informative slices of the sliced-Wasserstein distance reduces to one global constant, so the classical SWD with a tuned learning rate is competitive with specialized variants.