Towards Unsupervised Language Understanding and Generation by Joint Dual Learning

TL;DR

This paper proposes a flexible joint learning framework that exploits the duality between natural language understanding and generation, enabling both supervised and unsupervised training to improve performance in dialogue systems.

Contribution

It introduces a novel dual learning framework that effectively leverages the duality between NLU and NLG, allowing for combined supervised and unsupervised training methods.

Findings

Boosts performance of NLU and NLG components

Effective integration of supervised and unsupervised learning

Demonstrates improvements on benchmark datasets

Abstract

In modular dialogue systems, natural language understanding (NLU) and natural language generation (NLG) are two critical components, where NLU extracts the semantics from the given texts and NLG is to construct corresponding natural language sentences based on the input semantic representations. However, the dual property between understanding and generation has been rarely explored. The prior work is the first attempt that utilized the duality between NLU and NLG to improve the performance via a dual supervised learning framework. However, the prior work still learned both components in a supervised manner, instead, this paper introduces a general learning framework to effectively exploit such duality, providing flexibility of incorporating both supervised and unsupervised learning algorithms to train language understanding and generation models in a joint fashion. The benchmark experiments demonstrate that the proposed approach is capable of boosting the performance of both NLU and NLG.