AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

ProConMV: Provenance-Enabled Conceptual Framework for Interpretable Multi-View Diabetic Retinopathy Diagnosis

Xiaoling Luo, Shuo Yang, Qihao Xu, Jiansong Zhang, Zhuoqin Yang, Zhihui Lai, Linlin Shen, Chengliang Liu

ICML 2026 · Vision-language and multimodal › Human and embodied modeling · approach: Language-grounded visual concept learning

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 dual-uncertainty multi-view fusion (DU-MVFD) (ablation)ProConMV · MFIDDR (four views) ablationgrading accuracy: 86.23 -> 86.75Improved
add dual-uncertainty multi-view fusion (DU-MVFD) (ablation)ProConMV · MFIDDR (four views) ablationgrading kappa: 75.23 -> 76.05Improved
add text and VT-RWKV concept integration (ablation)ProConMV · MFIDDR (four views) ablationgrading accuracy: 85.68 -> 86.75Improved
add Hilbert-RWKV backbone (ablation)ProConMV · MFIDDR (four views) ablationgrading accuracy: 85.72 -> 86.75Improved
add lesion masks as input (ablation)ProConMV · MFIDDR (four views) ablationgrading accuracy: 85.21 -> 86.75Improved
add Hilbert-RWKV backbone to MVCBM baseline (ablation)MVCBM · MFIDDR (four views) ablationconcept F1: 59.10 -> 63.87Improved
add DU-MVFD alone to MVCBM baseline (ablation)MVCBM · MFIDDR (four views) ablationconcept AUPR: 61.56 -> 60.87Hurt
use Hilbert-RWKV instead of ResNet-50 backbone (ablation)ProConMV · MFIDDR (four views)concept AUPR: 67.71 -> 72.26Improved
use Hilbert-RWKV instead of ResNet-50 backbone (ablation)ProConMV · MFIDDR (four views)parameters and inference time: 25.26M -> 6.70M; 10.16ms -> 8.77msImproved
use Hilbert-RWKV instead of VMamba (ablation)ProConMV · MFIDDR (four views)grading accuracy: 83.03 -> 86.75Improved
use Hilbert scan instead of zigzag scan (ablation)Hilbert-RWKV · MFIDDR (four views) (scan strategy)concept F1: 63.89 -> 68.92Improved
use Hilbert scan instead of sweep scan (ablation)Hilbert-RWKV · MFIDDR (four views) (scan strategy)grading accuracy: 86.29 -> 86.75Improved

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