An NLP attack that swaps gradient-important words with synonyms while minimizing BERT embedding distance is reported to beat PWWS and BERT-Attack on accuracy drop, perturbation rate, and queries, but the paper gives no code or full setup.
Intriguing properties of neural networks
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CR 1years
2025 1verdicts
REJECT 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Adversarial Text Generation with Dynamic Contextual Perturbation
An NLP attack that swaps gradient-important words with synonyms while minimizing BERT embedding distance is reported to beat PWWS and BERT-Attack on accuracy drop, perturbation rate, and queries, but the paper gives no code or full setup.