Research map / Label-efficient and robust learning / Recognition and long-tail
Deep Hashing with Hash-Consistent Large Margin Proxy Embeddings
Pedro M. Morgado, Yunsheng Li, José Costa Pereira, Mohammad Javad Saberian, Nuno M. Vasconcelos
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 fixed sHCLM proxies instead of proxies learned by back-propagation (ablation) | proxy embedding hashing · CIFAR-100 (64 bits) | classification accuracy: 75.4 -> 75.5 | No clear effect |
| use fixed sHCLM proxies instead of learned proxies (ablation) | proxy embedding hashing · CIFAR-100 | retrieval mAP | Improved |
| rotate Tammes proxies to be more binary (Aligned) (ablation) | proxy embedding hashing · CIFAR-100 | retrieval mAP | Improved |
| assign proxies to classes semantically (sHCLM over HCLM) (ablation) | proxy embedding hashing · CIFAR-100 | retrieval mAP | Improved |
| add triplet loss to sHCLM (ablation) | sHCLM · CIFAR-10 (32 bits) | mAP: 95.2 -> 94.9 | No clear effect |
| add triplet loss to sHCLM (ablation) | sHCLM · CIFAR-100 (64 bits) | mAP: 78.0 -> 78.3 | No clear effect |
| add triplet loss to sHCLM (ablation) | sHCLM · ILSVRC-2012 (128 bits) | mAP: 32.4 -> 32.1 | No clear effect |
| add triplet loss to sHCLM (ablation) | sHCLM · CIFAR-100 transfer (16 bits) | mAP: 22.0 -> 27.0 | Improved |
| add triplet loss to sHCLM (ablation) | sHCLM · ILSVRC-2012 transfer (128 bits) | mAP: 17.4 -> 19.5 | Improved |
| add triplet loss to sHCLM (ablation) | sHCLM · CIFAR-100 transfer (64 bits) | classification accuracy: 65.0 -> 63.6 | Hurt |
| add triplet loss to sHCLM (ablation) | sHCLM · NUS-WIDE (48 bits) | mAP: 80.9 -> 83.5 | Improved |
| use sHCLM instead of SSDH proxy embedding | SSDH · ILSVRC-2012 (128 bits) | mAP: 28.9 -> 32.4 | Improved |
Closest papers by meaning
- Supervised Learning of Semantics-Preserving Hash via Deep Convolutional Neural Networks (TPAMI 2017 · similarity 0.89)
- Deep Supervised Hashing for Fast Image Retrieval (CVPR 2016 · similarity 0.89)
- Deep Hash Distillation for Image Retrieval (ECCV 2022 · similarity 0.88)
- From Hashing to CNNs: Training Binary Weight Networks via Hashing (AAAI 2018 · similarity 0.88)
- Greedy Hash: Towards Fast Optimization for Accurate Hash Coding in CNN (NeurIPS 2018 · similarity 0.88)
- Pairwise-Label-Based Deep Incremental Hashing with Simultaneous Code Expansion (AAAI 2024 · 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.
