AI Paper Atlas

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

Counterfactual Vision and Language Learning

Ehsan Abbasnejad, Damien Teney, Amin Parvaneh, Javen Shi, Anton van den Hengel

CVPR 2020 · Vision-language and multimodal › Visual question answering · approach: Vision-language model VQA evaluation · 130 citations (OpenAlex)

Check 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 counterfactual training on both questions and imagesUpDn · VQA-CP testOverall accuracy: 39.74 -> 42.12Improved
Add counterfactual training on both questions and imagesUpDn · VQA-CP testOther accuracy: 46.05 -> 48.34Improved
Intervene on questions only rather than both modalities (ablation)UpDn with counterfactual training · VQA-CP testNumber accuracy: 11.93 -> 14.17Improved
Intervene on images only rather than both modalities (ablation)UpDn with counterfactual training · VQA-CP testOverall accuracy: 42.12 -> 41.01Hurt
Intervene on questions only rather than both modalities (ablation)UpDn with counterfactual training · VQA-CP testOverall accuracy: 42.12 -> 40.62Hurt
Add counterfactual trainingPythia · VQA v2 validationOverall accuracy: 68.49 -> 68.77Improved
Add counterfactual trainingGRU · Embodied QA, T−10dT: 0.74 -> 0.67Improved
Add counterfactual trainingGRU · Embodied QA, T−50dΔ: 2.26 -> 2.52Improved
Use optimized counterfactual exogenous variables instead of posterior-only training (ablation)LSTM with randomly initialized embeddings · SSTAccuracy: 85.21 -> 85.61Improved
Use optimized counterfactual exogenous variables instead of posterior-only training (ablation)LSTM with trainable GloVe embeddings · SSTAccuracy: 87.24 -> 88.4Improved
Add counterfactual trainingVGG-19 · CIFAR-10Accuracy: 95.04 -> 96.73Improved
Add counterfactual trainingResNet-101 · CIFAR-100Accuracy: 77.78 -> 80.0Improved

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