Learning Task Sampling Policy for Multitask Learning

Dhanasekar Sundararaman; Henry Tsai; Kuang Huei Lee; Iulia Turc; Lawrence Carin

Learning Task Sampling Policy for Multitask Learning

Dhanasekar Sundararaman^*, Henry Tsai, Kuang Huei Lee, Iulia Turc, Lawrence Carin

^*Corresponding author for this work

Computer, Electrical and Mathematical Sciences and Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Scopus citations

Abstract

It has been shown that training multi-task models with auxiliary tasks can improve the target tasks quality through cross-task transfer. However, the importance of each auxiliary task to the primary task is likely not known a priori. While the importance weights of auxiliary tasks can be manually tuned, it becomes practically infeasible with the number of tasks scaling up. To address this, we propose a search method that automatically assigns importance weights. We formulate it as a reinforcement learning problem and learn a task sampling schedule based on evaluation accuracy of the multi-task model. Our empirical evaluation on XNLI and GLUE shows that our method outperforms uniform sampling and the corresponding single-task baseline.

Original language	English (US)
Title of host publication	Findings of the Association for Computational Linguistics, Findings of ACL
Subtitle of host publication	EMNLP 2021
Editors	Marie-Francine Moens, Xuanjing Huang, Lucia Specia, Scott Wen-Tau Yih
Publisher	Association for Computational Linguistics (ACL)
Pages	4410-4415
Number of pages	6
ISBN (Electronic)	9781955917100
State	Published - 2021
Event	2021 Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021 - Punta Cana, Dominican Republic Duration: Nov 7 2021 → Nov 11 2021

Publication series

Name	Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021

Conference

Conference	2021 Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021
Country/Territory	Dominican Republic
City	Punta Cana
Period	11/7/21 → 11/11/21

Bibliographical note

Publisher Copyright:
© 2021 Association for Computational Linguistics.

ASJC Scopus subject areas

Language and Linguistics
Linguistics and Language

Cite this

Sundararaman, D., Tsai, H., Lee, K. H., Turc, I., & Carin, L. (2021). Learning Task Sampling Policy for Multitask Learning. In M.-F. Moens, X. Huang, L. Specia, & S. W.-T. Yih (Eds.), Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021 (pp. 4410-4415). (Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021). Association for Computational Linguistics (ACL).

Sundararaman, Dhanasekar ; Tsai, Henry ; Lee, Kuang Huei et al. / Learning Task Sampling Policy for Multitask Learning. Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021. editor / Marie-Francine Moens ; Xuanjing Huang ; Lucia Specia ; Scott Wen-Tau Yih. Association for Computational Linguistics (ACL), 2021. pp. 4410-4415 (Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021).

@inproceedings{657480f3ee8b4217a6f44b5c0223cf7b,

title = "Learning Task Sampling Policy for Multitask Learning",

abstract = "It has been shown that training multi-task models with auxiliary tasks can improve the target tasks quality through cross-task transfer. However, the importance of each auxiliary task to the primary task is likely not known a priori. While the importance weights of auxiliary tasks can be manually tuned, it becomes practically infeasible with the number of tasks scaling up. To address this, we propose a search method that automatically assigns importance weights. We formulate it as a reinforcement learning problem and learn a task sampling schedule based on evaluation accuracy of the multi-task model. Our empirical evaluation on XNLI and GLUE shows that our method outperforms uniform sampling and the corresponding single-task baseline.",

author = "Dhanasekar Sundararaman and Henry Tsai and Lee, {Kuang Huei} and Iulia Turc and Lawrence Carin",

note = "Publisher Copyright: {\textcopyright} 2021 Association for Computational Linguistics.; 2021 Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021 ; Conference date: 07-11-2021 Through 11-11-2021",

year = "2021",

language = "English (US)",

series = "Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021",

publisher = "Association for Computational Linguistics (ACL)",

pages = "4410--4415",

editor = "Marie-Francine Moens and Xuanjing Huang and Lucia Specia and Yih, {Scott Wen-Tau}",

booktitle = "Findings of the Association for Computational Linguistics, Findings of ACL",

address = "United States",

}

Sundararaman, D, Tsai, H, Lee, KH, Turc, I & Carin, L 2021, Learning Task Sampling Policy for Multitask Learning. in M-F Moens, X Huang, L Specia & SW-T Yih (eds), Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021. Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021, Association for Computational Linguistics (ACL), pp. 4410-4415, 2021 Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021, Punta Cana, Dominican Republic, 11/7/21.

Learning Task Sampling Policy for Multitask Learning. / Sundararaman, Dhanasekar; Tsai, Henry; Lee, Kuang Huei et al.
Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021. ed. / Marie-Francine Moens; Xuanjing Huang; Lucia Specia; Scott Wen-Tau Yih. Association for Computational Linguistics (ACL), 2021. p. 4410-4415 (Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Learning Task Sampling Policy for Multitask Learning

AU - Sundararaman, Dhanasekar

AU - Tsai, Henry

AU - Lee, Kuang Huei

AU - Turc, Iulia

AU - Carin, Lawrence

PY - 2021

Y1 - 2021

N2 - It has been shown that training multi-task models with auxiliary tasks can improve the target tasks quality through cross-task transfer. However, the importance of each auxiliary task to the primary task is likely not known a priori. While the importance weights of auxiliary tasks can be manually tuned, it becomes practically infeasible with the number of tasks scaling up. To address this, we propose a search method that automatically assigns importance weights. We formulate it as a reinforcement learning problem and learn a task sampling schedule based on evaluation accuracy of the multi-task model. Our empirical evaluation on XNLI and GLUE shows that our method outperforms uniform sampling and the corresponding single-task baseline.

AB - It has been shown that training multi-task models with auxiliary tasks can improve the target tasks quality through cross-task transfer. However, the importance of each auxiliary task to the primary task is likely not known a priori. While the importance weights of auxiliary tasks can be manually tuned, it becomes practically infeasible with the number of tasks scaling up. To address this, we propose a search method that automatically assigns importance weights. We formulate it as a reinforcement learning problem and learn a task sampling schedule based on evaluation accuracy of the multi-task model. Our empirical evaluation on XNLI and GLUE shows that our method outperforms uniform sampling and the corresponding single-task baseline.

UR - http://www.scopus.com/inward/record.url?scp=85129176058&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85129176058

T3 - Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021

SP - 4410

EP - 4415

BT - Findings of the Association for Computational Linguistics, Findings of ACL

A2 - Moens, Marie-Francine

A2 - Huang, Xuanjing

A2 - Specia, Lucia

A2 - Yih, Scott Wen-Tau

PB - Association for Computational Linguistics (ACL)

T2 - 2021 Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021

Y2 - 7 November 2021 through 11 November 2021

ER -

Sundararaman D, Tsai H, Lee KH, Turc I, Carin L. Learning Task Sampling Policy for Multitask Learning. In Moens MF, Huang X, Specia L, Yih SWT, editors, Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021. Association for Computational Linguistics (ACL). 2021. p. 4410-4415. (Findings of the Association for Computational Linguistics, Findings of ACL: EMNLP 2021).

Learning Task Sampling Policy for Multitask Learning

Abstract

Publication series

Conference

Bibliographical note

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this