MetaASSIST: Robust Dialogue State Tracking with Meta Learning

TL;DR

MetaASSIST introduces a meta-learning framework to adaptively tune pseudo-label weighting in dialogue state tracking, significantly improving robustness and achieving state-of-the-art accuracy on MultiWOZ 2.4.

Contribution

It proposes a novel meta-learning approach to dynamically learn weighting parameters for pseudo-labels in DST, addressing limitations of fixed parameters.

Findings

Achieves a joint goal accuracy of 80.10% on MultiWOZ 2.4.

Demonstrates competitive performance with flexible weighting schemes.

Outperforms previous methods in robustness to noisy annotations.

Abstract

Existing dialogue datasets contain lots of noise in their state annotations. Such noise can hurt model training and ultimately lead to poor generalization performance. A general framework named ASSIST has recently been proposed to train robust dialogue state tracking (DST) models. It introduces an auxiliary model to generate pseudo labels for the noisy training set. These pseudo labels are combined with vanilla labels by a common fixed weighting parameter to train the primary DST model. Notwithstanding the improvements of ASSIST on DST, tuning the weighting parameter is challenging. Moreover, a single parameter shared by all slots and all instances may be suboptimal. To overcome these limitations, we propose a meta learning-based framework MetaASSIST to adaptively learn the weighting parameter. Specifically, we propose three schemes with varying degrees of flexibility, ranging from slot-wise to both slot-wise and instance-wise, to convert the weighting parameter into learnable functions. These functions are trained in a meta-learning manner by taking the validation set as meta data. Experimental results demonstrate that all three schemes can achieve competitive performance. Most impressively, we achieve a state-of-the-art joint goal accuracy of 80.10% on MultiWOZ 2.4.