Gated Task Interaction Framework for Multi-task Sequence Tagging

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Recent studies have shown that neural models can achieve high performance on several sequence labelling/tagging problems without the explicit use of linguistic features such as part-of-speech (POS) tags. These models are trained only using the character-level and the word embedding vectors as inputs. Others have shown that linguistic features can improve the performance of neural models on tasks such as chunking and named entity recognition (NER). However, the change in performance depends on the degree of semantic relatedness between the linguistic features and the target task; in some instances, linguistic features can have a negative impact on performance. This paper presents an approach to jointly learn
these linguistic features along with the target sequence labelling tasks with a new multi-task learning (MTL) framework called Gated Tasks Interaction (GTI) network for solving multiple sequence tagging tasks. The GTI network exploits the relations
between the multiple tasks via neural gate modules. These gate modules control the flow of information between the different tasks. Experiments on benchmark datasets for chunking and NER show that our framework outperforms other competitive baselines trained with and without external training resources.
LanguageEnglish
Title of host publication2019 International Joint Conference on Neural Networks, IJCNN 2019
Volume2019-July
ISBN (Electronic)978-1-7281-1985-4
DOIs
Publication statusAccepted/In press - 7 Mar 2019

Publication series

NameProceedings of International Joint Conference on Neural Networks
PublisherIEEE
ISSN (Print)2161-4393
ISSN (Electronic)2161-4407

Fingerprint

Linguistics
Labeling
Semantics
Experiments

Cite this

Ampomah, I., McClean, S. I., Lin, Z., & Hawe, G. (Accepted/In press). Gated Task Interaction Framework for Multi-task Sequence Tagging. In 2019 International Joint Conference on Neural Networks, IJCNN 2019 (Vol. 2019-July). [8851801] (Proceedings of International Joint Conference on Neural Networks ). https://doi.org/10.1109/IJCNN.2019.8851801
Ampomah, Isaac ; McClean, Sally I ; Lin, Zhiwei ; Hawe, Glenn. / Gated Task Interaction Framework for Multi-task Sequence Tagging. 2019 International Joint Conference on Neural Networks, IJCNN 2019. Vol. 2019-July 2019. (Proceedings of International Joint Conference on Neural Networks ).
@inproceedings{b6476425531244108acb1f98122dd5cf,
title = "Gated Task Interaction Framework for Multi-task Sequence Tagging",
abstract = "Recent studies have shown that neural models can achieve high performance on several sequence labelling/tagging problems without the explicit use of linguistic features such as part-of-speech (POS) tags. These models are trained only using the character-level and the word embedding vectors as inputs. Others have shown that linguistic features can improve the performance of neural models on tasks such as chunking and named entity recognition (NER). However, the change in performance depends on the degree of semantic relatedness between the linguistic features and the target task; in some instances, linguistic features can have a negative impact on performance. This paper presents an approach to jointly learnthese linguistic features along with the target sequence labelling tasks with a new multi-task learning (MTL) framework called Gated Tasks Interaction (GTI) network for solving multiple sequence tagging tasks. The GTI network exploits the relationsbetween the multiple tasks via neural gate modules. These gate modules control the flow of information between the different tasks. Experiments on benchmark datasets for chunking and NER show that our framework outperforms other competitive baselines trained with and without external training resources.",
author = "Isaac Ampomah and McClean, {Sally I} and Zhiwei Lin and Glenn Hawe",
year = "2019",
month = "3",
day = "7",
doi = "10.1109/IJCNN.2019.8851801",
language = "English",
isbn = "978-1-7281-1986-1",
volume = "2019-July",
series = "Proceedings of International Joint Conference on Neural Networks",
publisher = "IEEE",
booktitle = "2019 International Joint Conference on Neural Networks, IJCNN 2019",

}

Ampomah, I, McClean, SI, Lin, Z & Hawe, G 2019, Gated Task Interaction Framework for Multi-task Sequence Tagging. in 2019 International Joint Conference on Neural Networks, IJCNN 2019. vol. 2019-July, 8851801, Proceedings of International Joint Conference on Neural Networks . https://doi.org/10.1109/IJCNN.2019.8851801

Gated Task Interaction Framework for Multi-task Sequence Tagging. / Ampomah, Isaac; McClean, Sally I; Lin, Zhiwei; Hawe, Glenn.

2019 International Joint Conference on Neural Networks, IJCNN 2019. Vol. 2019-July 2019. 8851801 (Proceedings of International Joint Conference on Neural Networks ).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Gated Task Interaction Framework for Multi-task Sequence Tagging

AU - Ampomah, Isaac

AU - McClean, Sally I

AU - Lin, Zhiwei

AU - Hawe, Glenn

PY - 2019/3/7

Y1 - 2019/3/7

N2 - Recent studies have shown that neural models can achieve high performance on several sequence labelling/tagging problems without the explicit use of linguistic features such as part-of-speech (POS) tags. These models are trained only using the character-level and the word embedding vectors as inputs. Others have shown that linguistic features can improve the performance of neural models on tasks such as chunking and named entity recognition (NER). However, the change in performance depends on the degree of semantic relatedness between the linguistic features and the target task; in some instances, linguistic features can have a negative impact on performance. This paper presents an approach to jointly learnthese linguistic features along with the target sequence labelling tasks with a new multi-task learning (MTL) framework called Gated Tasks Interaction (GTI) network for solving multiple sequence tagging tasks. The GTI network exploits the relationsbetween the multiple tasks via neural gate modules. These gate modules control the flow of information between the different tasks. Experiments on benchmark datasets for chunking and NER show that our framework outperforms other competitive baselines trained with and without external training resources.

AB - Recent studies have shown that neural models can achieve high performance on several sequence labelling/tagging problems without the explicit use of linguistic features such as part-of-speech (POS) tags. These models are trained only using the character-level and the word embedding vectors as inputs. Others have shown that linguistic features can improve the performance of neural models on tasks such as chunking and named entity recognition (NER). However, the change in performance depends on the degree of semantic relatedness between the linguistic features and the target task; in some instances, linguistic features can have a negative impact on performance. This paper presents an approach to jointly learnthese linguistic features along with the target sequence labelling tasks with a new multi-task learning (MTL) framework called Gated Tasks Interaction (GTI) network for solving multiple sequence tagging tasks. The GTI network exploits the relationsbetween the multiple tasks via neural gate modules. These gate modules control the flow of information between the different tasks. Experiments on benchmark datasets for chunking and NER show that our framework outperforms other competitive baselines trained with and without external training resources.

UR - http://www.scopus.com/inward/record.url?scp=85073242311&partnerID=8YFLogxK

U2 - 10.1109/IJCNN.2019.8851801

DO - 10.1109/IJCNN.2019.8851801

M3 - Conference contribution

SN - 978-1-7281-1986-1

VL - 2019-July

T3 - Proceedings of International Joint Conference on Neural Networks

BT - 2019 International Joint Conference on Neural Networks, IJCNN 2019

ER -

Ampomah I, McClean SI, Lin Z, Hawe G. Gated Task Interaction Framework for Multi-task Sequence Tagging. In 2019 International Joint Conference on Neural Networks, IJCNN 2019. Vol. 2019-July. 2019. 8851801. (Proceedings of International Joint Conference on Neural Networks ). https://doi.org/10.1109/IJCNN.2019.8851801