Extend Adversarial Policy Against Neural Machine Translation via Unknown Token
This work addresses robustness issues in neural machine translation systems, but it is incremental as it builds on existing adversarial generation techniques.
The paper tackled the problem of adversarial example generation for neural machine translation by proposing a method that introduces character perturbations to overcome limitations of token-based adversarial policies, resulting in compatibility with scenarios where baseline methods fail and generation of high-efficiency adversarial examples.
Generating adversarial examples contributes to mainstream neural machine translation~(NMT) robustness. However, popular adversarial policies are apt for fixed tokenization, hindering its efficacy for common character perturbations involving versatile tokenization. Based on existing adversarial generation via reinforcement learning~(RL), we propose the `DexChar policy' that introduces character perturbations for the existing mainstream adversarial policy based on token substitution. Furthermore, we improve the self-supervised matching that provides feedback in RL to cater to the semantic constraints required during training adversaries. Experiments show that our method is compatible with the scenario where baseline adversaries fail, and can generate high-efficiency adversarial examples for analysis and optimization of the system.