Multi-Task Learning for Sequence Tagging: An Empirical Study

TL;DR

This paper empirically evaluates three multi-task learning approaches on 11 sequence tagging tasks, showing that joint learning often improves performance and reveals task relationships and clusters.

Contribution

It provides a comprehensive empirical analysis of MTL approaches on sequence tagging, highlighting when and how joint learning benefits tasks and uncovering task clusters.

Findings

Joint learning improves performance in about 50% of cases.

Pairwise MTL reveals beneficial and harmful task relationships.

One MTL approach uncovers natural clustering of semantic and syntactic tasks.

Abstract

We study three general multi-task learning (MTL) approaches on 11 sequence tagging tasks. Our extensive empirical results show that in about 50% of the cases, jointly learning all 11 tasks improves upon either independent or pairwise learning of the tasks. We also show that pairwise MTL can inform us what tasks can benefit others or what tasks can be benefited if they are learned jointly. In particular, we identify tasks that can always benefit others as well as tasks that can always be harmed by others. Interestingly, one of our MTL approaches yields embeddings of the tasks that reveal the natural clustering of semantic and syntactic tasks. Our inquiries have opened the doors to further utilization of MTL in NLP.