Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
RankDNN: Learning to Rank for Few-Shot Learning
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 |
|---|---|---|---|
| add RankDNN ranking to an IE ResNet-12 backbone | IE · miniImageNet 5-way 1-shot | accuracy: 67.28_{\pm 0.80} -> 68.72_{\pm 0.15} | Helped |
| add RankDNN ranking to an IE ResNet-12 backbone | IE · CUB 5-way 1-shot | accuracy: 80.92 -> 82.93 | Helped |
| fuse RankDNN with ProtoNet | ProtoNet · miniImageNet 5-way 1-shot | accuracy: 61.20 -> 64.75 (+3.55) | Helped |
| replace RankMLP with RankSVM (ablation) | S2M2 with Kronecker encoding · miniImageNet 5-way 1-shot | accuracy: 66.67 -> 60.1 | Hurt |
| replace Kronecker encoding with feature disparity (ablation) | IE RankDNN · miniImageNet 5-way 1-shot | accuracy: 68.72 -> 60.57 | Hurt |
| replace Kronecker encoding with concatenated triplet features (ablation) | IE RankDNN · miniImageNet 5-way 1-shot | accuracy: 68.72 -> 20 | Hurt |
| replace Kronecker encoding with pairwise-concatenation disparity (ablation) | IE RankDNN · miniImageNet 5-way 1-shot | accuracy: 68.72 -> 42.50 | Hurt |
| replace Kronecker encoding with Hadamard product (ablation) | IE RankDNN · miniImageNet 5-way 1-shot | accuracy: 68.72 -> 65.00 | Hurt |
| replace Kronecker encoding with Hadamard product (ablation) | IE RankDNN · CUB 5-way 1-shot | accuracy: 82.93 -> 81.64 | Helped |
| replace Kronecker encoding with polynomial product (ablation) | IE RankDNN · miniImageNet 5-way 1-shot | accuracy: 68.72 -> 21.00 | Hurt |
| combine Kronecker and Hadamard encodings (ablation) | S2M2 RankDNN · miniImageNet 5-way 1-shot | accuracy: 66.67 -> 66.88 | Helped |
| apply PCA dimension reduction to IE features (ablation) | IE baseline · miniImageNet 5-way 1-shot | accuracy: 67.15 -> 65.32 | Hurt |
Closest papers by meaning
- Re-ranking for image retrieval and transductive few-shot classification (NeurIPS 2021 · similarity 0.91)
- RankMI: A Mutual Information Maximizing Ranking Loss (CVPR 2020 · similarity 0.89)
- Few-Shot Learning Through an Information Retrieval Lens (NeurIPS 2017 · similarity 0.87)
- Learning to Compare: Relation Network for Few-Shot Learning (CVPR 2018 · similarity 0.87)
- SoDeep: A Sorting Deep Net to Learn Ranking Loss Surrogates (CVPR 2019 · similarity 0.86)
- NUDGE: Lightweight Non-Parametric Fine-Tuning of Embeddings for Retrieval (ICLR 2025 · similarity 0.84)
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.