Cross-lingual Distillation for Text Classification

TL;DR

This paper introduces a cross-lingual text classification method using model distillation and adversarial feature adaptation, enabling effective classification in languages without labeled data, and demonstrates competitive results on benchmark datasets.

Contribution

It presents a novel cross-lingual distillation framework with adversarial adaptation for text classification in low-resource languages.

Findings

Achieved comparable performance to state-of-the-art methods

Effective in languages with no labeled training data

Validated on multiple benchmark datasets

Abstract

Cross-lingual text classification(CLTC) is the task of classifying documents written in different languages into the same taxonomy of categories. This paper presents a novel approach to CLTC that builds on model distillation, which adapts and extends a framework originally proposed for model compression. Using soft probabilistic predictions for the documents in a label-rich language as the (induced) supervisory labels in a parallel corpus of documents, we train classifiers successfully for new languages in which labeled training data are not available. An adversarial feature adaptation technique is also applied during the model training to reduce distribution mismatch. We conducted experiments on two benchmark CLTC datasets, treating English as the source language and German, French, Japan and Chinese as the unlabeled target languages. The proposed approach had the advantageous or comparable performance of the other state-of-art methods.