Curriculum Learning Effectively Improves Low Data VQA

Narjes Askarian, Ehsan Abbasnejad, Ingrid Zukerman, Wray Buntine, Gholamreza Haffari


Abstract
Visual question answering (VQA) models, in particular modular ones, are commonly trained on large-scale datasets to achieve state-of-the-art performance. However, such datasets are sometimes not available. Further, it has been shown that training these models on small datasets significantly reduces their accuracy. In this paper, we propose curriculum-based learning (CL) regime to increase the accuracy of VQA models trained on small datasets. Specifically, we offer three criteria to rank the samples in these datasets and propose a training strategy for each criterion. Our results show that, for small datasets, our CL approach yields more accurate results than those obtained when training with no curriculum.
Anthology ID:
2021.alta-1.3
Volume:
Proceedings of the 19th Annual Workshop of the Australasian Language Technology Association
Month:
December
Year:
2021
Address:
Online
Editors:
Afshin Rahimi, William Lane, Guido Zuccon
Venue:
ALTA
SIG:
Publisher:
Australasian Language Technology Association
Note:
Pages:
22–33
Language:
URL:
https://aclanthology.org/2021.alta-1.3
DOI:
Bibkey:
Cite (ACL):
Narjes Askarian, Ehsan Abbasnejad, Ingrid Zukerman, Wray Buntine, and Gholamreza Haffari. 2021. Curriculum Learning Effectively Improves Low Data VQA. In Proceedings of the 19th Annual Workshop of the Australasian Language Technology Association, pages 22–33, Online. Australasian Language Technology Association.
Cite (Informal):
Curriculum Learning Effectively Improves Low Data VQA (Askarian et al., ALTA 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.alta-1.3.pdf
Data
CLEVRVisual Question Answering