CG-BERT: Conditional Text Generation with BERT for Generalized Few-shot Intent Detection
This addresses a realistic challenge in natural language understanding for intent detection systems, offering a novel method to handle few-shot scenarios, though it is incremental in building upon pre-trained models like BERT.
The paper tackles the problem of Generalized Few-Shot Intent Detection (GFSID), which involves discriminating both existing and novel intents with limited data, by proposing CG-BERT, a model that generates text conditioned on intent labels using BERT and variational inference, achieving state-of-the-art performance with 1-shot and 5-shot settings on two datasets.
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.