Prompt Conditioned VAE: Enhancing Generative Replay for Lifelong Learning in Task-Oriented Dialogue

TL;DR

This paper introduces Prompt Conditioned VAE (PCLL), a novel generative replay method that uses task-specific prompts and distillation to improve lifelong learning in task-oriented dialogue systems, effectively reducing catastrophic forgetting.

Contribution

PCLL enhances generative replay by incorporating task statistics via prompts and distillation, outperforming existing methods in lifelong learning for dialogue systems.

Findings

PCLL significantly outperforms baseline methods in natural language understanding tasks.

Incorporating task-specific prompts improves pseudo-sample quality.

Distillation helps in consolidating past knowledge effectively.

Abstract

Lifelong learning (LL) is vital for advanced task-oriented dialogue (ToD) systems. To address the catastrophic forgetting issue of LL, generative replay methods are widely employed to consolidate past knowledge with generated pseudo samples. However, most existing generative replay methods use only a single task-specific token to control their models. This scheme is usually not strong enough to constrain the generative model due to insufficient information involved. In this paper, we propose a novel method, prompt conditioned VAE for lifelong learning (PCLL), to enhance generative replay by incorporating tasks' statistics. PCLL captures task-specific distributions with a conditional variational autoencoder, conditioned on natural language prompts to guide the pseudo-sample generation. Moreover, it leverages a distillation process to further consolidate past knowledge by alleviating the noise in pseudo samples. Experiments on natural language understanding tasks of ToD systems demonstrate that PCLL significantly outperforms competitive baselines in building LL models.