This repository contains code for the paper:
- Ignorance is Bliss: Exploring Defenses Against Invariance-based Attacks on Neural Machine Translation Systems. IEEE TAI, 2021 [pdf]
This code is for two contrasting defense strategies against invariance-based attack for NMT systems. The two defenese strategies, namely, Learn to Deal (LTD) and Learn to Ignore (LTI), perform noise-aware training on NMT systems.
The repo for Min-Grad + Soft-Att attack is [here].