Prior Omission of Dissimilar Source Domain(s) for Cost-Effective Few-Shot Learning

TL;DR

This paper introduces a similarity-based data selection method and a Shared-Private Network for few-shot slot tagging, effectively reducing negative transfer from dissimilar source domains and improving adaptation performance.

Contribution

It proposes a novel similarity-based source data selection technique and a Shared-Private Network architecture for more effective few-shot domain adaptation in slot tagging.

Findings

Outperforms state-of-the-art methods with fewer source data

Redundant data from dissimilar sources can harm adaptation

Shared features improve label prediction in limited data scenarios

Abstract

Few-shot slot tagging is an emerging research topic in the field of Natural Language Understanding (NLU). With sufficient annotated data from source domains, the key challenge is how to train and adapt the model to another target domain which only has few labels. Conventional few-shot approaches use all the data from the source domains without considering inter-domain relations and implicitly assume each sample in the domain contributes equally. However, our experiments show that the data distribution bias among different domains will significantly affect the adaption performance. Moreover, transferring knowledge from dissimilar domains will even introduce some extra noises so that affect the performance of models. To tackle this problem, we propose an effective similarity-based method to select data from the source domains. In addition, we propose a Shared-Private Network (SP-Net) for the few-shot slot tagging task. The words from the same class would have some shared features. We extract those shared features from the limited annotated data on the target domain and merge them together as the label embedding to help us predict other unlabelled data on the target domain. The experiment shows that our method outperforms the state-of-the-art approaches with fewer source data. The result also proves that some training data from dissimilar sources are redundant and even negative for the adaption.