Effectiveness of Pre-training for Few-shot Intent Classification

TL;DR

This paper demonstrates that fine-tuning BERT with a small labeled dataset is highly effective for few-shot intent classification, outperforming more complex pre-training methods on diverse domains.

Contribution

It introduces IntentBERT, a simple fine-tuning approach that achieves superior few-shot intent classification performance across various domains.

Findings

Fine-tuning BERT with ~1,000 labeled examples outperforms existing pre-trained models.

IntentBERT generalizes well across different domains with minimal labeled data.

Simple fine-tuning is highly effective for few-shot intent classification.

Abstract

This paper investigates the effectiveness of pre-training for few-shot intent classification. While existing paradigms commonly further pre-train language models such as BERT on a vast amount of unlabeled corpus, we find it highly effective and efficient to simply fine-tune BERT with a small set of labeled utterances from public datasets. Specifically, fine-tuning BERT with roughly 1,000 labeled data yields a pre-trained model -- IntentBERT, which can easily surpass the performance of existing pre-trained models for few-shot intent classification on novel domains with very different semantics. The high effectiveness of IntentBERT confirms the feasibility and practicality of few-shot intent detection, and its high generalization ability across different domains suggests that intent classification tasks may share a similar underlying structure, which can be efficiently learned from a small set of labeled data. The source code can be found at https://github.com/hdzhang-code/IntentBERT.