Research map / Vision-language and multimodal / Cross-modal learning
Aggregating Image and Text Quantized Correlated Components
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Aggregate both modalities using MACC instead of using early-fused or single-modality features | Pascal VOC07 classification · Pascal VOC07 | mAP: 86.16 -> 90.12 | Helped |
| Aggregate KCCA projections into MACC representations rather than directly using KCCA(VGG+W2V) | Flickr 8K text-to-image retrieval · Flickr 8K | R@1: 26.1 -> 27.6 | Helped |
| Complete the missing modality with nearest-neighbor information rather than leaving it incomplete (ablation) | MACC cross-modal classification · Pascal VOC07 Text-Image | mAP: 12.03 -> 82.17 | Helped |
| Complete the missing modality with nearest-neighbor information rather than leaving it incomplete (ablation) | MACC cross-modal classification · Pascal VOC07 Image-Text | mAP: 10.04 -> 79.18 | Helped |
| Use soft or local soft coding instead of hard coding (ablation) | MACC classification · Pascal VOC07 | mAP: 88.75 -> 90.37 | Helped |
| Increase the auxiliary completion dataset from Flickr 8K training data to Flickr 30K training data (ablation) | MACC retrieval · Flickr 30K | R@1: 33.9 -> 35.3 | Helped |
| Use codebook size k=16 and n=5 nearest codewords rather than k=8 and n=1 (ablation) | MACC classification · Pascal VOC07 | mAP: 88.75 -> 90.37 | Helped |
| Use different codebook sizes and coding neighborhood sizes (ablation) | MACC classification · Pascal VOC07 | mAP: 87.73; 89.71; 90.37; 89.68 | Mixed |
Closest papers by meaning
- Identifiable Shared Component Analysis of Unpaired Multimodal Mixtures (NeurIPS 2024 · similarity 0.87)
- Cross the Gap: Exposing the Intra-modal Misalignment in CLIP via Modality Inversion (ICLR 2025 · similarity 0.86)
- Connect, Collapse, Corrupt: Learning Cross-Modal Tasks with Uni-Modal Data (ICLR 2024 · similarity 0.86)
- On the Value of Cross-Modal Misalignment in Multimodal Representation Learning (NeurIPS 2025 · similarity 0.85)
- Accept the Modality Gap: An Exploration in the Hyperbolic Space (CVPR 2024 · similarity 0.85)
- When Kernels Multiply, Clusters Unify: Fusing Embeddings with the Kronecker Product (NeurIPS 2025 · similarity 0.85)
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.