A dynamic ensemble of BERT, RoBERTa, and ALBERT with randomized smoothing, masked inference, and TextFooler adversarial training is reported to keep 82-87% accuracy under attack on AG News and IMDB, far above prior defenses.
Saliency attention and semantic similarity-driven adversarial perturbation,
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
2024 1verdicts
REJECT 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Adversarial Robustness through Dynamic Ensemble Learning
A dynamic ensemble of BERT, RoBERTa, and ALBERT with randomized smoothing, masked inference, and TextFooler adversarial training is reported to keep 82-87% accuracy under attack on AG News and IMDB, far above prior defenses.