SOrT-ing VQA Models : Contrastive Gradient Learning for Improved Consistency

Sameer Dharur, Purva Tendulkar, Dhruv Batra, Devi Parikh, Ramprasaath R. Selvaraju


Abstract
Recent research in Visual Question Answering (VQA) has revealed state-of-the-art models to be inconsistent in their understanding of the world - they answer seemingly difficult questions requiring reasoning correctly but get simpler associated sub-questions wrong. These sub-questions pertain to lower level visual concepts in the image that models ideally should understand to be able to answer the reasoning question correctly. To address this, we first present a gradient-based interpretability approach to determine the questions most strongly correlated with the reasoning question on an image, and use this to evaluate VQA models on their ability to identify the relevant sub-questions needed to answer a reasoning question. Next, we propose a contrastive gradient learning based approach called Sub-question Oriented Tuning (SOrT) which encourages models to rank relevant sub-questions higher than irrelevant questions for an <image, reasoning-question> pair. We show that SOrT improves model consistency by up to 6.5% points over existing approaches, while also improving visual grounding and robustness to rephrasings of questions.
Anthology ID:
2021.naacl-main.248
Volume:
Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies
Month:
June
Year:
2021
Address:
Online
Editors:
Kristina Toutanova, Anna Rumshisky, Luke Zettlemoyer, Dilek Hakkani-Tur, Iz Beltagy, Steven Bethard, Ryan Cotterell, Tanmoy Chakraborty, Yichao Zhou
Venue:
NAACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
3103–3111
Language:
URL:
https://aclanthology.org/2021.naacl-main.248
DOI:
10.18653/v1/2021.naacl-main.248
Bibkey:
Cite (ACL):
Sameer Dharur, Purva Tendulkar, Dhruv Batra, Devi Parikh, and Ramprasaath R. Selvaraju. 2021. SOrT-ing VQA Models : Contrastive Gradient Learning for Improved Consistency. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 3103–3111, Online. Association for Computational Linguistics.
Cite (Informal):
SOrT-ing VQA Models : Contrastive Gradient Learning for Improved Consistency (Dharur et al., NAACL 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.naacl-main.248.pdf
Optional supplementary data:
 2021.naacl-main.248.OptionalSupplementaryData.zip
Optional supplementary code:
 2021.naacl-main.248.OptionalSupplementaryCode.zip
Video:
 https://aclanthology.org/2021.naacl-main.248.mp4
Code
 sameerdharur/sorting-vqa
Data
VQA-HATVisual Question Answering