Research map / Vision-language and multimodal / Cross-modal learning
Variational Adapter for Cross-modal Similarity Representation
WenZhang Wei, Zhipeng Gui, Dehua Peng, Tiandi Ye, Huayi Wu
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 |
|---|---|---|---|
| use VACSR instead of PCME++ | CLIP ViT-B/32 · COCO 5K image-to-text | R@1: 62.1 -> 66.5 | Improved |
| use VACSR instead of PCME++ | CLIP ViT-B/16 · ECCV Caption image-to-text | mAP@R: 34.5 -> 35.4 | Improved |
| use VACSR instead of NPC at 20% noise | CLIP ViT-B/32 · COCO noisy correspondence | RSUM: 529.8 -> 539.0 | Improved |
| use VACSR under 50% noise versus next-best | CLIP ViT-B/32 · COCO noisy correspondence | RSUM: 523.4 -> 534.2 | Improved |
| use VACSR instead of MMRL | CLIP ViT-B/16 · ImageNet OOD average | accuracy: 60.6 -> 61.6 | Improved |
| use VACSR instead of CoPrompt on novel classes | CLIP ViT-B/16 · Base-to-novel 11 datasets | novel accuracy: 77.23 -> 76.08 | Hurt |
| use VACSR instead of MMRL on harmonic mean | CLIP ViT-B/16 · Base-to-novel 11 datasets | harmonic mean: 81.20 -> 80.37 | Hurt |
| add Gaussian mixture (GMM) prior instead of single Gaussian (ablation) | VACSR · ECCV Caption | mAP@R: 39.6 -> 40.7 | Improved |
| add distributional optimization loss L_sigma (ablation) | VACSR · ECCV Caption | mAP@R: 39.8 -> 40.7 | Improved |
| add L_sigma (ablation) | VACSR · CxC | R@1: 60.7 -> 59.7 | Hurt |
| add KL regularization (ablation) | VACSR · ECCV Caption | mAP@R: 40.1 -> 40.7 | Improved |
| add sigmoid normalization of variance (ablation) | VACSR · ECCV Caption | mAP@R: 40.1 -> 40.7 | 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.
- MMRL — MMRL: Multi-Modal Representation Learning for Vision-Language Models (CVPR 2025) · better on 2 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
- Cross the Gap: Exposing the Intra-modal Misalignment in CLIP via Modality Inversion (ICLR 2025 · similarity 0.90)
- WikiDO: A New Benchmark Evaluating Cross-Modal Retrieval for Vision-Language Models (NeurIPS 2024 · similarity 0.90)
- Robust Noisy Correspondence Learning with Equivariant Similarity Consistency (CVPR 2024 · similarity 0.90)
- Dynamic Adapter with Semantics Disentangling for Cross-lingual Cross-modal Retrieval (AAAI 2025 · similarity 0.89)
- Same Concept, Different Directions: Cross-Modal Feature Heterogeneity in Sparse Autoencoders (NeurIPS 2026 · similarity 0.89)
- Noisy Correspondence Rectification via Asymmetric Similarity Learning (AAAI 2025 · similarity 0.88)
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.
