AI Paper Atlas

Research map / Vision-language and multimodal / Visual question answering

VQA-LOL: Visual Question Answering under the Lens of Logic

Tejas Gokhale, Pratyay Banerjee, Chitta Baral, Yezhou Yang

ECCV 2020 · Vision-language and multimodal › Visual question answering · approach: Universal VQA models · 84 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add logically composed training data (VQA-Compose)LXMERT · VQA-Compose valaccuracy (%): 50.79 -> 85.03Improved
add logically composed training data (VQA-Compose)LXMERT · VQA valaccuracy (%): 68.94 -> 67.85Hurt
add Supplement data to Compose trainingLXMERT · VQA-Compose valaccuracy (%): 85.03 -> 70.28Hurt
replace LXMERT with LOL attention modules (qATT) (ablation)LXMERT · VQA-Compose valaccuracy (%): 70.28 -> 79.88Improved
use full LOL (question + logic attention) instead of qATT only (ablation)LOL · VQA-Compose valaccuracy (%): 79.88 -> 81.07Improved
add Frechet-Compatibility loss to LOL (Full) (ablation)LOL (Full) · VQA-Compose valaccuracy (%): 81.07 -> 82.39Improved
add Frechet-Compatibility loss to LOL (Full) (ablation)LOL (Full) · VQA valaccuracy (%): 68.86 -> 68.10Hurt
add Frechet-Compatibility loss to LXMERT (ablation)LXMERT · VQA-Compose valaccuracy (%): 70.28 -> 75.31Improved
use LOL instead of LXMERT for generalization from single to multiple connectivesLOL · VQA-Compose val (multiple connectives)accuracy (%): 61.99 -> 66.03Improved
use end-to-end LOL without a parser instead of LOL with trained parser (ablation)LOL (full) · VQA-Compose + Supplement val overallaccuracy (%): 83.15 -> 85.10Improved
use trained parser vs oracle parser with LOL (ablation)LOL (full) · VQA-Compose + Supplement val overallaccuracy (%): 87.15 -> 83.15Hurt
train LOL on closed questions only instead of LXMERTLOL (lATT) · VQA-Supplement val (closed-question training)accuracy (%): 88.15 -> 89.00Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy