Few-shot Learning for Multi-label Intent Detection

TL;DR

This paper proposes a novel few-shot learning approach for multi-label intent detection, utilizing universal thresholding and label name embeddings to improve accuracy with limited data.

Contribution

It introduces a universal thresholding method and label name embedding technique to enhance multi-label intent detection in few-shot scenarios.

Findings

Significant performance improvement over baselines in one-shot and five-shot settings.

Effective threshold calibration using nonparametric learning.

Enhanced class separation through label name embeddings.

Abstract

In this paper, we study the few-shot multi-label classification for user intent detection. For multi-label intent detection, state-of-the-art work estimates label-instance relevance scores and uses a threshold to select multiple associated intent labels. To determine appropriate thresholds with only a few examples, we first learn universal thresholding experience on data-rich domains, and then adapt the thresholds to certain few-shot domains with a calibration based on nonparametric learning. For better calculation of label-instance relevance score, we introduce label name embedding as anchor points in representation space, which refines representations of different classes to be well-separated from each other. Experiments on two datasets show that the proposed model significantly outperforms strong baselines in both one-shot and five-shot settings.