A vector-quantized autoencoder with parameter-disentangled streams and phonological semi-supervision reconstructs and identifies unseen ASL signs better than a standard VQ-VAE on the Sem-Lex benchmark.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Phonological Representation Learning for Isolated Signs Improves Out-of-Vocabulary Generalization
A vector-quantized autoencoder with parameter-disentangled streams and phonological semi-supervision reconstructs and identifies unseen ASL signs better than a standard VQ-VAE on the Sem-Lex benchmark.