AI Paper Atlas

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

Taking a HINT: Leveraging Explanations to Make Vision and Language Models More Grounded

Ramprasaath R. Selvaraju, Stefan Lee, Yilin Shen, Hongxia Jin, Shalini Ghosh, Larry Heck, Dhruv Batra, Devi Parikh

ICCV 2019 · Vision-language and multimodal › Visual question answering · approach: Vision-language model VQA evaluation · 263 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 HINT gradient-importance fine-tuningUpDn · VQA-CP testoverall accuracy: 39.49 -> 46.73Improved
add HINT fine-tuningUpDn · VQAv2 valoverall accuracy: 62.85 -> 63.38Improved
supervise attention weights (Attn. Align) instead of gradient importance (ablation)UpDn · VQA-CP testoverall accuracy: 39.49 -> 39.37No clear effect
supervise attention weights (Attn. Align) (ablation)UpDn · VQAv2 valoverall accuracy: 62.85 -> 63.24Improved
use HINT instead of adversarial regularizationUpDn · VQA-CP testoverall accuracy: 41.17 -> 46.73Improved
add HINTUpDn · VQA-CP testNumber accuracy: 11.:96 -> 10.61Hurt
use 1.5% of VQA data with human attention supervisionUpDn + HINT · VQA-CP testoverall accuracy: 5% improvementImproved
add HINTBottom-up Top-down captioner · COCO captioning (robust split)CIDEr: 0.90 -> 0.92Improved
add HINTBottom-up Top-down captioner · COCO captioning (Karpathy split)CIDEr: 1.06 -> 1.04Hurt
add HINTUpDn · VQA-CPSpearman correlation with human attention: -0.09 -> 0.18Improved
add HINTBottom-up Top-down captioner · COCO robust splitSpearman correlation with segmentation maps: 0.008 -> 0.17Improved
add HINTUpDn · VQA-CPIoU of top proposal with human attention: 0.57 -> 0.63Improved

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