AI Paper Atlas

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

Human Uncertainty-Aware Data Selection and Automatic Labeling in Visual Question Answering

Jian Lan, Zhicheng Liu, Udo Schlegel, Raoyuan Zhao, Yihong Liu, Hinrich Schuetze, Michael A. Hedderich, Thomas Seidl

ICLR 2026 · Vision-language and multimodal › Visual question answering · approach: Vision-language model VQA evaluation

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 HU-aware Selectors (vs random samples) (ablation)HaDola · VQAv2, Qwen2.5VL-7BHU-acc: 72.23 -> 76.75Improved
add Error Trigger (ablation)HaDola · VQAv2, Qwen2.5VL-7BHU-acc: 71.56 -> 76.75Improved
use self-annotation instead of human labels (ablation)HaDola · VQAv2, Qwen2.5VL-7BHU-acc: 73.91 -> 76.75Improved
use HaDola loss instead of cross-entropy (ablation)HaDola · VQAv2, Qwen2.5VL-7BHU-acc: 72.37 -> 76.75Improved
train on high-HU samples vs low-HU samplesSFT · VQAv2HU-accHurt
add high-HU data to low+medium HU trainingSFT · VQAv2HU-accHurt
combine low- and medium-HU samplesSFT · VQAv2HU-accImproved
use HU-acc instead of VQA-acc to evaluate subsetsEvaluation metric · VQAv2metric separationImproved
HaDola with 5% labels vs SFT with 100% annotationsHaDola · VQAv2/VizWiz, zero-shot VLMsHU-accImproved
HaDola on BEiT3HaDola · VQAv2/VizWiz, BEiT3HU-accNo clear effect
increase labeled fraction beyond 15%HaDola · VQAv2, VizWizHU-accNo clear effect
change (beta, lambda) weights (ablation)HaDola · VQAv2, Qwen2.5VL-7BHU-acc: 76.75 -> 68.43 (worst)Mixed

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