Research map / Label-efficient and robust learning / Self-supervised and contrastive
Self-Supervised Adversarial Hashing Networks for Cross-Modal Retrieval
Read the paperCheck 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 |
|---|---|---|---|
| Use SSAH instead of the strongest CNN-F baseline | Cross-modal retrieval · Flickr-25K, I→T, 16 bits | MAP: 0.735 -> 0.782 | Helped |
| Use SSAH instead of the strongest CNN-F baseline | Cross-modal retrieval · Flickr-25K, T→I, 16 bits | MAP: 0.763 -> 0.791 | Helped |
| Use SSAH instead of the strongest CNN-F baseline | Cross-modal retrieval · NUS-WIDE, I→T, 16 bits | MAP: 0.533 -> 0.642 | Helped |
| Use SSAH instead of the strongest CNN-F baseline | Cross-modal retrieval · NUS-WIDE, T→I, 16 bits | MAP: 0.638 -> 0.669 | Helped |
| Use SSAH instead of the strongest CNN-F baseline | Cross-modal retrieval · MS COCO, I→T, 16 bits | MAP: 0.483 -> 0.550 | Helped |
| Use SSAH instead of the strongest CNN-F baseline | Cross-modal retrieval · MS COCO, T→I, 16 bits | MAP: 0.501 -> 0.537 | Helped |
| Use SSAH instead of ACMR | Cross-modal retrieval · NUS-WIDE-10k, I→T | MAP: 0.544 -> 0.617 | Helped |
| Use SSAH instead of ACMR | Cross-modal retrieval · NUS-WIDE-10k, T→I | MAP: 0.538 -> 0.641 | Helped |
| Add the self-supervised semantic network (LabNet) (ablation) | SSAH · MIRFLICKR-25K, 16 bits | retrieval performance | Helped |
| Use the multi-scale text fusion model instead of three fully connected layers (ablation) | SSAH · MIRFLICKR-25K, 16 bits | retrieval performance | Helped |
| Add adversarial learning (ablation) | SSAH · MIRFLICKR-25K, 16 bits | retrieval performance | Helped |
| Use SSAH rather than DCMH | Cross-modal hashing training · Training efficiency | training speed: a factor of 10 | Helped |
Closest papers by meaning
- Unsupervised Generative Adversarial Cross-Modal Hashing (AAAI 2018 · similarity 0.91)
- Cross-Modal Learning with Adversarial Samples (NeurIPS 2019 · similarity 0.90)
- Robust Self-Paced Hashing for Cross-Modal Retrieval with Noisy Labels (AAAI 2025 · similarity 0.90)
- Dual Self-Paced Cross-Modal Hashing (AAAI 2024 · similarity 0.89)
- An End-To-End Graph Attention Network Hashing for Cross-Modal Retrieval (NeurIPS 2024 · similarity 0.87)
- Lightweight Contrastive Distilled Hashing for Online Cross-modal Retrieval (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.