On the ADReSS benchmark, a Ridge model reaches 87.8% accuracy in leave-one-subject-out validation and an extreme minimal learning machine reaches 79.2% on the held-out test set, both with fewer than 200 acoustic features.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.AS 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Dementia classification from spontaneous speech using wrapper-based feature selection
On the ADReSS benchmark, a Ridge model reaches 87.8% accuracy in leave-one-subject-out validation and an extreme minimal learning machine reaches 79.2% on the held-out test set, both with fewer than 200 acoustic features.