AdvAug: Robust Adversarial Augmentation for Neural Machine Translation

TL;DR

AdvAug introduces a novel adversarial augmentation technique for neural machine translation that enhances model robustness and translation quality by utilizing virtual sentences in a smooth interpolated embedding space, outperforming existing methods.

Contribution

The paper presents a new adversarial augmentation method, AdvAug, which leverages a novel vicinity distribution for virtual sentences to improve NMT performance without extra data.

Findings

AdvAug improves BLEU scores by up to 4.9 points on multiple benchmarks.

It outperforms traditional data augmentation techniques like back-translation.

The method enhances model robustness and translation quality.

Abstract

In this paper, we propose a new adversarial augmentation method for Neural Machine Translation (NMT). The main idea is to minimize the vicinal risk over virtual sentences sampled from two vicinity distributions, of which the crucial one is a novel vicinity distribution for adversarial sentences that describes a smooth interpolated embedding space centered around observed training sentence pairs. We then discuss our approach, AdvAug, to train NMT models using the embeddings of virtual sentences in sequence-to-sequence learning. Experiments on Chinese-English, English-French, and English-German translation benchmarks show that AdvAug achieves significant improvements over the Transformer (up to 4.9 BLEU points), and substantially outperforms other data augmentation techniques (e.g. back-translation) without using extra corpora.