AI Paper Atlas

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

Reliable Visual Question Answering: Abstain Rather Than Answer Incorrectly

Spencer Whitehead, Suzanne Petryk, Vedaad Shakib, Joseph Gonzalez, Trevor Darrell, Anna Rohrbach, Marcus Rohrbach

ECCV 2022 · Vision-language and multimodal › Visual question answering · approach: Universal VQA models · 50 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 learned Selector instead of MaxProbCLIP-ViL · VQA v2 testCoverage at 1% risk: 6.83 -> 15.66Improved
add learned Selector instead of MaxProbPythia · VQA v2 testCoverage at 1% risk: 5.84 -> 8.30Improved
add vector-scaling calibration to MaxProbCLIP-ViL · VQA v2 testCoverage at 1% risk: 6.83 -> 12.43Improved
add Selector with high wrong-answer cost (c=100) instead of MaxProbCLIP-ViL · VQA v2 testEffective Reliability Phi_100: 2.82 -> 8.00Improved
allow abstention (Selector) instead of always answering at c=100CLIP-ViL · VQA v2 testEffective Reliability Phi_100: -2162.82 -> 8.00Improved
add multimodal input to Selector rather than only question (ablation)Selector on CLIP-ViL · VQA v2 selection validationCoverage at 1% risk: 0.02 -> 12.92Improved
regress VQA accuracy instead of classifying correctness (ablation)Selector on CLIP-ViL · VQA v2 selection validationCoverage at 1% risk: 6.51 -> 12.92Improved
use 2-layer MLP instead of 1-layer linear Selector (ablation)Selector on CLIP-ViL · VQA v2 selection validationCoverage at 1% risk: 10.38 -> 12.92Improved
use 4-layer Transformer instead of 2-layer MLP Selector (ablation)Selector on CLIP-ViL · VQA v2 selection validationCoverage at 1% risk: 12.92 -> 13.25No clear effect
use extra Dev data to augment VQA training (MaxProb-Aug) instead of training Selector (ablation)CLIP-ViL · VQA v2 selection validationCoverage at 1% risk: 12.92 -> 6.57Hurt
augment VQA training data with Dev set for MaxProbCLIP-ViL MaxProb · VQA v2 selection validationVQA accuracy: 69.70 -> 70.52Improved

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