Continual Generalized Intent Discovery: Marching Towards Dynamic and Open-world Intent Recognition

Xiaoshuai Song, Yutao Mou, Keqing He, Yueyan Qiu, Pei Wang +1 more
10/16/2023
cs.CLcs.AIcs.LG

No official code found. Best implementation: songxiaoshuai/CGID (4 stars, MIT, last commit 2024-02-21).

Abstract

In a practical dialogue system, users may input out-of-domain (OOD) queries. The Generalized Intent Discovery (GID) task aims to discover OOD intents from OOD queries and extend them to the in-domain (IND) classifier. However, GID only considers one stage of OOD learning, and needs to utilize the data in all previous stages for joint training, which limits its wide application in reality. In this paper, we introduce a new task, Continual Generalized Intent Discovery (CGID), which aims to continuously and automatically discover OOD intents from dynamic OOD data streams and then incrementally add them to the classifier with almost no previous data, thus moving towards dynamic intent recognition in an open world. Next, we propose a method called Prototype-guided Learning with Replay and Distillation (PLRD) for CGID, which bootstraps new intent discovery through class prototypes and balances new and old intents through data replay and feature distillation. Finally, we conduct detailed experiments and analysis to verify the effectiveness of PLRD and understand the key challenges of CGID for future research.

View on arXivView PDF

Code Implementations(1)

songxiaoshuai/CGIDCommunity100%
MIT

Official implementation of "Continual Generalized Intent Discovery: Marching Towards Dynamic and Open-world Intent Recognition", EMNLP23 Findings.

40Shell, PythonOct 11, 20232 years agoMIT

Similar Papers

Large Language Models Meet Open-World Intent Discovery and Recognition: An Evaluation of ChatGPT

85% match2023

Out-of-Domain Intent Detection Considering Multi-Turn Dialogue Contexts

82% match2023

Integration of Old and New Knowledge for Generalized Intent Discovery: A Consistency-driven Prototype-Prompting Framework

82% match1 repo62025

Decoupling Pseudo Label Disambiguation and Representation Learning for Generalized Intent Discovery

81% match2023

Selective In-Context Data Augmentation for Intent Detection using Pointwise V-Information

81% match2023

Novel Intent Detection and Active Learning Based Classification (Student Abstract)

80% match2023

Cite this paper

@article{song2023continual,
  title  = {Continual Generalized Intent Discovery: Marching Towards Dynamic and Open-world Intent Recognition},
  author = {Xiaoshuai Song and Yutao Mou and Keqing He and Yueyan Qiu and Pei Wang and Weiran Xu},
  year   = {2023},
  eprint = {2310.10184},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2310.10184v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2310.10184.svg)](https://codeofpaper.com/paper/2310.10184)

Discussion