CG-BERT: Conditional Text Generation with BERT for Generalized Few-shot Intent Detection

TL;DR

This paper introduces CG-BERT, a model that uses BERT for conditional text generation to improve generalized few-shot intent detection, effectively handling both existing and novel intents with limited data.

Contribution

The paper proposes CG-BERT, a novel approach leveraging BERT and variational inference for effective few-shot intent detection in a generalized setting.

Findings

Achieves state-of-the-art results on GFSID with 1-shot and 5-shot data.

Effectively generates diverse utterances for novel intents.

Outperforms existing methods on real-world datasets.

Abstract

In this paper, we formulate a more realistic and difficult problem setup for the intent detection task in natural language understanding, namely Generalized Few-Shot Intent Detection (GFSID). GFSID aims to discriminate a joint label space consisting of both existing intents which have enough labeled data and novel intents which only have a few examples for each class. To approach this problem, we propose a novel model, Conditional Text Generation with BERT (CG-BERT). CG-BERT effectively leverages a large pre-trained language model to generate text conditioned on the intent label. By modeling the utterance distribution with variational inference, CG-BERT can generate diverse utterances for the novel intents even with only a few utterances available. Experimental results show that CG-BERT achieves state-of-the-art performance on the GFSID task with 1-shot and 5-shot settings on two real-world datasets.