Effects of data and entity ablation on multitask learning models for biomedical entity recognition
Effects of data and entity ablation on multitask learning models for biomedical entity recognition
复制标题
数据和实体消融对生物医学实体识别多任务学习模型的影响
DOI：
10.1016/j.jbi.2022.104062
复制
发表时间：
2022
影响因子：
4.5
通讯作者：
McInnes, Bridget T.
中科院分区：
文献类型：
--
作者：
Rodriguez, Nicholas E.;Nguyen, Mai;McInnes, Bridget T.
动机训练特定于领域的命名实体识别（NER）模型需要高质量的手工策划的黄金标准数据集，这些数据集创建起来既耗时又昂贵。此外，当任务数量很大时，部署NLP模型所需的存储和内存可能会令人望而却步。在这项工作中，我们探索利用多任务学习来减少训练新的特定领域模型所需的训练数据量。我们评估我们的系统在22个不同的生物医学NER数据集，并评估在何种程度上迁移学习有助于任务性能使用两种形式的ablation.ResultsWe发现，多任务模型一般不提高性能，但在许多情况下执行标准杆相比，单任务模型。然而，我们表明，在某些情况下，新的看不见的任务可以通过从多任务模型的权重开始，使用更少的数据作为单个模型进行训练，并提高性能。https://github.com/NLPatVCU/multitasking_bert-1
MotivationTraining domain-specific named entity recognition (NER) models requires high quality hand curated gold standard datasets which are time-consuming and expensive to create. Furthermore, the storage and memory required to deploy NLP models can be prohibitive when the number of tasks is large. In this work, we explore utilizing multi-task learning to reduce the amount of training data needed to train new domain-specific models. We evaluate our system across 22 distinct biomedical NER datasets and evaluate the extent to which transfer learning helps task performance using two forms of ablation.ResultsWe found that multitasking models generally do not improve performance, but in many cases perform on par compared to single-task models. However, we show that in some cases, new unseen tasks can be trained as a single model using less data by starting with weights from a multitask model and improve performance.AvailabilityThe software underlying this article are available in: https://github.com/NLPatVCU/multitasking_bert-1.