AI Paper Atlas

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

Visual Question Answering as a Meta Learning Task

Damien Teney, Anton van den Hengel

ECCV 2018 · Vision-language and multimodal › Visual question answering · approach: Universal VQA models · 44 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 dynamic weights (ablation)2 static prototypes, L2 · VQA-Numbersaverage answer recall: 31.48 -> 31.81Improved
add dynamic prototypes on top of dynamic weights (ablation)dynamic-weights model · VQA-Numbersaverage answer recall: 31.81 -> 32.32Improved
swap dot product for L2 distance (ablation)1 static prototype per answer · VQA-Numbersaverage answer recall: 29.72 -> 27.80Hurt
swap dot product for L1 similarity (ablation)1 static prototype per answer · VQA-Numbersaverage answer recall: 29.72 -> 29.97No clear effect
use two static prototypes per answer with L2 similarity (ablation)dot-product baseline · VQA-Numbersaverage answer recall: 29.72 -> 31.48Improved
use two static prototypes with L1 similarity (ablation)2 static prototypes dot product · VQA-Numbersaverage answer recall: 30.28 -> 28.34Hurt
add dynamic prototypes only (ablation)VQA v2 baseline (100% data) · VQA v2 valanswer recall: 15.8 -> 18.0Improved
add dynamic weights only (ablation)VQA v2 baseline (100% data) · VQA v2 valanswer recall: 15.8 -> 16.3Improved
add dynamic weights and prototypes together (ablation)VQA v2 baseline (50% data) · VQA v2 valanswer recall: 14.0 -> 15.5Improved
use proposed model instead of baselineVQA v2 baseline (100% data) · VQA v2 valquestion accuracy: 59.8 -> 59.9No clear effect
remove static prototypes, keep only dynamic (ablation)dynamic prototypes model (100% data) · VQA v2 valanswer recall: 18.0 -> 4.29Hurt
train on only two of seven answers with full support setproposed model · VQA-Numbersaverage answer recall: 14.28 -> 17.05Improved

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