Gradient-based Adversarial Attacks against Text Transformers
Adversarial examples are not perturbed, but sampled from a distribution
#adversarial-attack
#natural-language-processing
BBAEG: Towards BERT-based Biomedical Adversarial Example Generation for Text Classification
Doctors might think that they have heard of the anti-psychotic drug 'clozapinum' somewhere
#adversarial-attack
#natural-language-processing