Finding a Needle in the Adversarial Haystack: A Targeted Paraphrasing Approach For Uncovering Edge Cases with Minimal Distribution Distortion

Adversarial attacks against language models(LMs) are a significant concern. In particular, adversarial samples exploit the model's sensitivity to small input changes. While these changes appear insignificant on the semantics of the input sample, they result in significant decay in model performance. In this paper, we propose Targeted Paraphrasing via RL (TPRL), an approach to automatically learn a policy to generate challenging samples that most likely improve the model's performance. TPRL leverages FLAN T5, a language model, as a generator and …

adversarial adversarial attacks attacks cases cs.cl distribution edge exploit haystack language language models lms paraphrasing performance sample samples semantics sensitivity small