Randomized smoothing in audio classification requires explicit specification of the certified representation and preprocessing because different choices produce different certified accuracies and effective perturbation scales even at identical noise levels on keyword spotting and environmental sound
arXiv preprint arXiv:2404.18791 , year=
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.AS 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Representation Matters in Randomized Smoothing for Audio Classification
Randomized smoothing in audio classification requires explicit specification of the certified representation and preprocessing because different choices produce different certified accuracies and effective perturbation scales even at identical noise levels on keyword spotting and environmental sound