Label-aware Multi-level Contrastive Learning for Cross-lingual Spoken Language Understanding

TL;DR

This paper introduces a multi-level contrastive learning framework with label-awareness and novel code-switching schemes to improve zero-shot cross-lingual spoken language understanding, explicitly modeling hierarchical semantic structures.

Contribution

It proposes a novel multi-level contrastive learning approach with label-awareness and new code-switching schemes to better align semantic structures across languages in SLU.

Findings

Significant performance improvements over strong baselines.

Effective explicit alignment of utterance, slot, and word levels.

Enhanced zero-shot cross-lingual SLU accuracy.

Abstract

Despite the great success of spoken language understanding (SLU) in high-resource languages, it remains challenging in low-resource languages mainly due to the lack of labeled training data. The recent multilingual code-switching approach achieves better alignments of model representations across languages by constructing a mixed-language context in zero-shot cross-lingual SLU. However, current code-switching methods are limited to implicit alignment and disregard the inherent semantic structure in SLU, i.e., the hierarchical inclusion of utterances, slots, and words. In this paper, we propose to model the utterance-slot-word structure by a multi-level contrastive learning framework at the utterance, slot, and word levels to facilitate explicit alignment. Novel code-switching schemes are introduced to generate hard negative examples for our contrastive learning framework. Furthermore, we develop a label-aware joint model leveraging label semantics to enhance the implicit alignment and feed to contrastive learning. Our experimental results show that our proposed methods significantly improve the performance compared with the strong baselines on two zero-shot cross-lingual SLU benchmark datasets.