AI Paper Atlas

Research map / Vision-language and multimodal / Cross-modal learning

Decoupling Common and Unique Representations for Multimodal Self-supervised Learning

Yi Wang, Conrad M Albrecht, Nassim Ait Ali Braham, Chenying Liu, Zhitong Xiong, Xiao Xiang Zhu

ECCV 2024 · Vision-language and multimodal › Cross-modal learning · approach: Multimodal fusion with missing modalities · 80 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 intra-modal training and modality decoupling to cross-modal Barlow Twins (ablation)DeCUR · GeoNRW RGB-DEMmIoU: 33.6 -> 36.9Improved
add modality decoupling alone without intra-modal training (ablation)DeCUR · SAR-optical and RGB-DEMmAP / mIoU: 80.3 -> 80.1 (SAR-optical); 33.6 -> 34.3 (RGB-DEM)Mixed
add intra-modal training on top of decoupling (ablation)DeCUR · SAR-opticalmAP: 80.1 -> 81.7Improved
add decoupling on top of intra-modal training (ablation)DeCUR · GeoNRW RGB-DEMmIoU: 35.2 -> 36.9Improved
add deformable attention without residual connection (ablation)DeCUR · label-limited frozen settingsmAP: -0.1 to -1.2 at 1% labelsHurt
add residual deformable attention (RDA) (ablation)DeCUR · BigEarthNet and GeoNRW frozenmAP / mIoU: +0.4 to +2.4 in 7 of 8 settingsImproved
use cross-modal deformable attention instead of RDA (ablation)DeCUR · BigEarthNet-SAR 1%mAP: 74.4 -> 72.5Hurt
replace RDA with CBAM (ablation)DeCUR · BigEarthNet-SAR 1%mAP: 74.4 -> 73.1Hurt
decouple up to 50% of embedding dimensions as unique (ablation)DeCUR · SAR-optical and RGB-DEMdownstream performanceNo clear effect
swap cross-modal SimCLR for DeCUR with ViT-S/16ViT backbone · BigEarthNet-SAR 1% frozenmAP: 73.9 -> 75.8Improved
pretrain with DeCUR instead of single-modal Barlow TwinsSAR-only transfer · BigEarthNet SAR-only 1% labelsmAP: 71.2/73.3 -> 74.4/76.0Improved

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