End-to-End Joint Learning of Natural Language Understanding and Dialogue Manager

TL;DR

This paper introduces an end-to-end deep learning model that jointly learns natural language understanding and dialogue management, significantly improving performance over traditional pipeline approaches by reducing error propagation.

Contribution

The paper presents a novel joint training approach for NLU and dialogue policy learning, enhancing robustness and accuracy in conversational systems.

Findings

Outperforms state-of-the-art pipeline models in NLU and SAP tasks

Mitigates effects of noisy NLU outputs through joint training

Refines NLU model via backpropagation from system action signals

Abstract

Natural language understanding and dialogue policy learning are both essential in conversational systems that predict the next system actions in response to a current user utterance. Conventional approaches aggregate separate models of natural language understanding (NLU) and system action prediction (SAP) as a pipeline that is sensitive to noisy outputs of error-prone NLU. To address the issues, we propose an end-to-end deep recurrent neural network with limited contextual dialogue memory by jointly training NLU and SAP on DSTC4 multi-domain human-human dialogues. Experiments show that our proposed model significantly outperforms the state-of-the-art pipeline models for both NLU and SAP, which indicates that our joint model is capable of mitigating the affects of noisy NLU outputs, and NLU model can be refined by error flows backpropagating from the extra supervised signals of system actions.