Multi-Task Learning for Sequence Tagging: An Empirical Study
This empirical study provides insights into task interactions for NLP researchers, but it is incremental as it builds on existing multi-task learning methods without introducing new paradigms.
The authors investigated three multi-task learning approaches across 11 sequence tagging tasks, finding that joint learning improved performance in about 50% of cases compared to independent or pairwise learning, and identified tasks that consistently benefit or harm others.
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.