HuCurl: Human-induced Curriculum Discovery

Mohamed Elgaar, Hadi Amiri


Abstract
We introduce the problem of curriculum discovery and describe a curriculum learning framework capable of discovering effective curricula in a curriculum space based on prior knowledge about sample difficulty. Using annotation entropy and loss as measures of difficulty, we show that (i): the top-performing discovered curricula for a given model and dataset are often non-monotonic as apposed to monotonic curricula in existing literature, (ii): the prevailing easy-to-hard or hard-to-easy transition curricula are often at the risk of underperforming, and (iii): the curricula discovered for smaller datasets and models perform well on larger datasets and models respectively. The proposed framework encompasses some of the existing curriculum learning approaches and can discover curricula that outperform them across several NLP tasks.
Anthology ID:
2023.acl-long.104
Volume:
Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
Month:
July
Year:
2023
Address:
Toronto, Canada
Editors:
Anna Rogers, Jordan Boyd-Graber, Naoaki Okazaki
Venue:
ACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
1862–1877
Language:
URL:
https://aclanthology.org/2023.acl-long.104
DOI:
10.18653/v1/2023.acl-long.104
Bibkey:
Cite (ACL):
Mohamed Elgaar and Hadi Amiri. 2023. HuCurl: Human-induced Curriculum Discovery. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1862–1877, Toronto, Canada. Association for Computational Linguistics.
Cite (Informal):
HuCurl: Human-induced Curriculum Discovery (Elgaar & Amiri, ACL 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.acl-long.104.pdf
Video:
 https://aclanthology.org/2023.acl-long.104.mp4