AI Paper Atlas

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

Learning Unseen Modality Interaction

Yunhua Zhang, Hazel Doughty, Cees G. M. Snoek

NeurIPS 2023 · Vision-language and multimodal › Cross-modal learning · approach: Multimodal fusion with missing modalities · 1 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 feature projection into a common space (ablation)vanilla multimodal transformer · Video classificationtop-1 accuracy (%): 17.2 -> 21.3Improved
Add dual branch prediction with pseudo-supervision (ablation)transformer with feature projection · Video classificationtop-1 accuracy (%): 21.3 -> 23.7Improved
Use vanilla transformer instead of simple late fusion averaging (ablation)multimodal fusion · Video classificationtop-1 accuracy (%): 18.0 -> 17.2Hurt
Add feature alignment loss L_align (ablation)full model · Video classificationtop-1 accuracy (%): 20.2 -> 21.4Improved
Add pseudo-supervised loss L_pseudo (ablation)full model with L_align · Video classificationtop-1 accuracy (%): 21.4 -> 23.7Improved
Reduce parameters from 14.4M to 7.2M to match vanilla transformer (ablation)full model · EPIC-Kitchenstop-1 accuracy (%): 23.7 -> 21.4Mixed
Replace feature projection with average pool plus linear layer (ablation)full model · EPIC-Kitchens RGB+Flow+Audiotop-1 accuracy (%): 23.7 -> 19.0Hurt
Add noise N(0,1) to one modality, compare to vanilla transformerfull model · Video classificationtop-1 accuracy (%): 10.1 -> 18.0Improved
Vary feature token length k* (ablation)feature projection · Video classificationtop-1 accuracy (%): 20.9 -> 23.7 -> 21.2Mixed
Vary number of learnable tokens n_u from 32 to 128 (ablation)feature alignment loss · Multimedia retrievalMnR: 73.2 -> 66.2Improved
Change pseudo to supervised token split from 50:50 to 30:70 or 70:30 (ablation)dual branch prediction · Video classificationtop-1 accuracy (%): 23.7 -> 22.9Hurt
Vary alpha from 0 to 3000 (ablation)pseudo loss weight · Video classificationtop-1 accuracy (%): 21.4 -> 23.7Improved

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