Improving Zero-Shot Cross-Lingual Transfer Learning via Robust Training

TL;DR

This paper introduces a robust training approach using adversarial training and randomized smoothing to enhance zero-shot cross-lingual transfer in multilingual models, especially for low-resource languages, without relying on costly parallel corpora.

Contribution

It proposes a novel robust training strategy that improves cross-lingual transfer by making models tolerant to embedding noise, reducing dependence on language alignment data.

Findings

Robust training significantly improves zero-shot transfer performance.

Enhanced results in generalized cross-lingual transfer with mixed-language inputs.

Robust methods outperform standard fine-tuning in low-resource scenarios.

Abstract

Pre-trained multilingual language encoders, such as multilingual BERT and XLM-R, show great potential for zero-shot cross-lingual transfer. However, these multilingual encoders do not precisely align words and phrases across languages. Especially, learning alignments in the multilingual embedding space usually requires sentence-level or word-level parallel corpora, which are expensive to be obtained for low-resource languages. An alternative is to make the multilingual encoders more robust; when fine-tuning the encoder using downstream task, we train the encoder to tolerate noise in the contextual embedding spaces such that even if the representations of different languages are not aligned well, the model can still achieve good performance on zero-shot cross-lingual transfer. In this work, we propose a learning strategy for training robust models by drawing connections between adversarial examples and the failure cases of zero-shot cross-lingual transfer. We adopt two widely used robust training methods, adversarial training and randomized smoothing, to train the desired robust model. The experimental results demonstrate that robust training improves zero-shot cross-lingual transfer on text classification tasks. The improvement is more significant in the generalized cross-lingual transfer setting, where the pair of input sentences belong to two different languages.