Composed Variational Natural Language Generation for Few-shot Intents

TL;DR

This paper introduces CLANG, a transformer-based variational autoencoder that generates training examples for few-shot intents by modeling domain and action components, improving intent detection in imbalanced scenarios.

Contribution

Proposes a novel composed variational generator (CLANG) that effectively models intent components for few-shot learning, with contrastive regularization enhancing generation quality.

Findings

Achieves state-of-the-art results on intent detection datasets.

Effectively models intent components for better few-shot learning.

Improves generation quality with contrastive regularization.

Abstract

In this paper, we focus on generating training examples for few-shot intents in the realistic imbalanced scenario. To build connections between existing many-shot intents and few-shot intents, we consider an intent as a combination of a domain and an action, and propose a composed variational natural language generator (CLANG), a transformer-based conditional variational autoencoder. CLANG utilizes two latent variables to represent the utterances corresponding to two different independent parts (domain and action) in the intent, and the latent variables are composed together to generate natural examples. Additionally, to improve the generator learning, we adopt the contrastive regularization loss that contrasts the in-class with the out-of-class utterance generation given the intent. To evaluate the quality of the generated utterances, experiments are conducted on the generalized few-shot intent detection task. Empirical results show that our proposed model achieves state-of-the-art performances on two real-world intent detection datasets.