Exploring Zero and Few-shot Techniques for Intent Classification
This addresses storage and resource constraints for NLU providers scaling to many customers, but it is incremental as it builds on existing methods like T-few and Flan-T5.
The paper tackles the cold-start problem in scaling intent classification for conversational NLU by exploring zero and few-shot techniques, finding that parameter-efficient fine-tuning with T-few on Flan-T5 yields the best performance even with one sample per intent.
Conversational NLU providers often need to scale to thousands of intent-classification models where new customers often face the cold-start problem. Scaling to so many customers puts a constraint on storage space as well. In this paper, we explore four different zero and few-shot intent classification approaches with this low-resource constraint: 1) domain adaptation, 2) data augmentation, 3) zero-shot intent classification using descriptions large language models (LLMs), and 4) parameter-efficient fine-tuning of instruction-finetuned language models. Our results show that all these approaches are effective to different degrees in low-resource settings. Parameter-efficient fine-tuning using T-few recipe (Liu et al., 2022) on Flan-T5 (Chang et al., 2022) yields the best performance even with just one sample per intent. We also show that the zero-shot method of prompting LLMs using intent descriptions