Research map / Label-efficient and robust learning / Recognition and long-tail
Local Similarity-Aware Deep Feature Embedding
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 position-dependent PDDM instead of feature-difference-only LMDM | Jointly trained metric and quadruplet embedding · CUB-200-2011 test | Recall@1 (%): 49.5 -> 55.0 | Helped |
| Use position-dependent PDDM instead of feature-difference-only LMDM | Jointly trained metric and quadruplet embedding · CARS196 test | Recall@1 (%): 50.9 -> 55.2 | Helped |
| Use PDDM+Quadruplet instead of LiftedStruct | Image retrieval embedding · CUB-200-2011 test | Recall@1 (%): 47.2 -> 58.3 | Helped |
| Use PDDM+Quadruplet instead of LiftedStruct | Image retrieval embedding · CARS196 test | Recall@1 (%): 49.0 -> 57.4 | Helped |
| Use PDDM+Quadruplet instead of PDDM+Triplet | PDDM-based image retrieval embedding · CUB-200-2011 test | Recall@1 (%): 50.9 -> 58.3 | Helped |
| Use PDDM+Quadruplet instead of PDDM+Triplet | PDDM-based image retrieval embedding · CARS196 test | Recall@1 (%): 46.4 -> 57.4 | Helped |
| Use PDDM hard mining with triplet embedding instead of the Triplet baseline | Triplet embedding · CUB-200-2011 test | Recall@1 (%): 36.1 -> 50.9 | Helped |
| Use PDDM hard mining with triplet embedding instead of the Triplet baseline | Triplet embedding · CARS196 test | Recall@1 (%): 39.1 -> 46.4 | Helped |
| Add feature-level embedding loss E_e (ablation) | PDDM score retrieval · Image retrieval | Recall@1 (%): 3.4%+ | Helped |
| Add feature-level embedding loss E_e (ablation) | PDDM+Quadruplet retrieval · Image retrieval | Recall@1 (%): 6.5%+ | Helped |
| Reduce mini-batch size from 64 to 32 (ablation) | PDDM+Quadruplet · CUB-200-2011 test | Recall@1 (%): 55.7% | Hurt |
| Increase mini-batch size to 132 (ablation) | PDDM+Quadruplet · Image retrieval | Recall@1 (%) | No clear effect |
Closest papers by meaning
- Divide and Conquer the Embedding Space for Metric Learning (CVPR 2019 · similarity 0.89)
- Deep Metric Learning via Lifted Structured Feature Embedding (CVPR 2016 · similarity 0.88)
- Signal-To-Noise Ratio: A Robust Distance Metric for Deep Metric Learning (CVPR 2019 · similarity 0.88)
- Stabilizing Sample Similarity in Representation via Mitigating Random Consistency (ICML 2025 · similarity 0.86)
- Volume-Aware Distance for Robust Similarity Learning (ICML 2025 · similarity 0.86)
- Euclidean Embedding of Data Using Local Distances (NeurIPS 2026 · 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.