Learning to Bridge Metric Spaces: Few-shot Joint Learning of Intent Detection and Slot Filling

TL;DR

This paper introduces ConProm, a novel few-shot learning method that jointly learns intent detection and slot filling by bridging metric spaces, significantly improving performance on dialogue understanding tasks with limited data.

Contribution

Proposes a contrastive prototype merging network that bridges metric spaces for joint intent and slot learning in few-shot settings, a novel approach in dialogue understanding.

Findings

Outperforms strong baselines on Snips and FewJoint datasets.

Effective in one-shot and five-shot learning scenarios.

Successfully bridges intent and slot metric spaces for improved joint learning.

Abstract

In this paper, we investigate few-shot joint learning for dialogue language understanding. Most existing few-shot models learn a single task each time with only a few examples. However, dialogue language understanding contains two closely related tasks, i.e., intent detection and slot filling, and often benefits from jointly learning the two tasks. This calls for new few-shot learning techniques that are able to capture task relations from only a few examples and jointly learn multiple tasks. To achieve this, we propose a similarity-based few-shot learning scheme, named Contrastive Prototype Merging network (ConProm), that learns to bridge metric spaces of intent and slot on data-rich domains, and then adapt the bridged metric space to the specific few-shot domain. Experiments on two public datasets, Snips and FewJoint, show that our model significantly outperforms the strong baselines in one and five shots settings.