Research map / Label-efficient and robust learning / Semi-supervised learning
Towards Multimodal Domain Generalization with Few Labels
Hongzhao Li, Hao Dong, Hualei Wan, Shupan Li, Mingliang Xu, Muhammad Haris Khan
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add CDCR, DAR and CMPA to the weakly supervised baseline (ablation) | SSMDG baseline · HAC, 5 labels per class | top-1 accuracy (%): 45.39 -> 60.77 | Improved |
| add CMPA on top of CDCR + DAR (ablation) | SSMDG framework · HAC, 5 labels per class | top-1 accuracy (%): 58.64 -> 60.77 | Improved |
| add CDCR and DAR together instead of either alone (ablation) | SSMDG framework · HAC, 5 labels per class | top-1 accuracy (%): 54.76 / 54.89 -> 58.64 | Improved |
| anchor consensus on fused prediction supported by one modality instead of averaging modality predictions (ablation) | CDCR · HAC, 5 labels per class | top-1 accuracy (%): 58.26 -> 60.77 | Improved |
| use fused-anchored consensus instead of requiring all views to agree (ablation) | CDCR · HAC, 5 labels per class | top-1 accuracy (%): 59.16 -> 60.77 | Improved |
| use generalized cross-entropy instead of standard cross-entropy on non-consensus samples (ablation) | DAR · HAC, 5 labels per class | top-1 accuracy (%): 58.32 -> 60.77 | Improved |
| apply DAR to both weak and strong views instead of one (ablation) | DAR · HAC, 5 labels per class | top-1 accuracy (%): 60.42 / 60.73 -> 60.77 | Improved |
| add cross-domain prototype alignment (ablation) | CMPA · HAC, 5 labels per class | top-1 accuracy (%): 58.34 -> 60.77 | Improved |
| add cross-modal translation to prototype alignment (ablation) | CMPA · HAC, 5 labels per class | top-1 accuracy (%): 60.75 -> 60.77 | Improved |
| align both weak and strong views instead of one (ablation) | CMPA · HAC, 5 labels per class | top-1 accuracy (%): 59.38 -> 60.77 | Improved |
| synthesize missing video via cross-modal translation instead of zero-filling (ablation) | SSMDG framework at test time · HAC, 5% labels per class | top-1 accuracy (%): 28.27 -> 35.86 | Improved |
| synthesize missing audio via cross-modal translation instead of zero-filling (ablation) | SSMDG framework at test time · HAC, 5% labels per class | top-1 accuracy (%): 52.45 -> 55.03 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- STiL — STiL: Semi-supervised Tabular-Image Learning for Comprehensive Task-Relevant Information Exploration in Multimodal Classification (CVPR 2025) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Multimodal Unsupervised Domain Generalization by Retrieving Across the Modality Gap (ICLR 2025 · similarity 0.90)
- Are We Making Progress in Multimodal Domain Generalization? A Comprehensive Benchmark Study (NeurIPS 2026 · similarity 0.90)
- Towards Generalizing to Unseen Domains with Few Labels (CVPR 2024 · similarity 0.88)
- Reasoning-Driven Multimodal LLM for Domain Generalization (ICLR 2026 · similarity 0.88)
- With Limited Data for Multimodal Alignment, Let the STRUCTURE Guide You (NeurIPS 2025 · similarity 0.88)
- Semi-Supervised Multimodal Classification Through Learning from Modal and Strategic Complementarities (AAAI 2025 · similarity 0.87)
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.
