Robust Textual Embedding against Word-level Adversarial Attacks
This addresses the problem of adversarial robustness in NLP for practitioners, though it is incremental as it builds on existing metric learning techniques.
The paper tackles the vulnerability of NLP models to word-level adversarial attacks by proposing Fast Triplet Metric Learning (FTML), a robust training method that significantly improves model robustness against various attacks while maintaining competitive classification accuracy on original samples.
We attribute the vulnerability of natural language processing models to the fact that similar inputs are converted to dissimilar representations in the embedding space, leading to inconsistent outputs, and we propose a novel robust training method, termed Fast Triplet Metric Learning (FTML). Specifically, we argue that the original sample should have similar representation with its adversarial counterparts and distinguish its representation from other samples for better robustness. To this end, we adopt the triplet metric learning into the standard training to pull words closer to their positive samples (i.e., synonyms) and push away their negative samples (i.e., non-synonyms) in the embedding space. Extensive experiments demonstrate that FTML can significantly promote the model robustness against various advanced adversarial attacks while keeping competitive classification accuracy on original samples. Besides, our method is efficient as it only needs to adjust the embedding and introduces very little overhead on the standard training. Our work shows great potential of improving the textual robustness through robust word embedding.