Randomized Smoothing with Masked Inference for Adversarially Robust Text Classifications
This addresses the brittleness of large-scale pre-trained language models against adversarial attacks, offering a practical method for robust training, though it is incremental as it builds on existing techniques.
The paper tackles the problem of adversarial robustness in NLP systems by introducing RSMI, a two-stage framework combining randomized smoothing and masked inference, which improves adversarial robustness by 2 to 3 times over state-of-the-art methods on benchmark datasets.
Large-scale pre-trained language models have shown outstanding performance in a variety of NLP tasks. However, they are also known to be significantly brittle against specifically crafted adversarial examples, leading to increasing interest in probing the adversarial robustness of NLP systems. We introduce RSMI, a novel two-stage framework that combines randomized smoothing (RS) with masked inference (MI) to improve the adversarial robustness of NLP systems. RS transforms a classifier into a smoothed classifier to obtain robust representations, whereas MI forces a model to exploit the surrounding context of a masked token in an input sequence. RSMI improves adversarial robustness by 2 to 3 times over existing state-of-the-art methods on benchmark datasets. We also perform in-depth qualitative analysis to validate the effectiveness of the different stages of RSMI and probe the impact of its components through extensive ablations. By empirically proving the stability of RSMI, we put it forward as a practical method to robustly train large-scale NLP models. Our code and datasets are available at https://github.com/Han8931/rsmi_nlp