Research map / Architectures and efficiency / RNNs and normalization
Shortlist Selection With Residual-Aware Distance Estimator for K-Nearest Neighbor Search
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 |
|---|---|---|---|
| Add residual-aware shortlist selection to the inverted index | II · GIST384-80M, K=100, shortlist 1M | recall: 0.837 -> 0.890 | Helped |
| Add residual-aware shortlist selection to the inverted multi-index | IMI · GIST384-80M, K=100, shortlist 1M | recall: 0.949 -> 0.976 | Helped |
| Add residual-aware shortlist selection to the inverted index | II · SIFT-1B, K=100, shortlist 1M | recall: 0.872 -> 0.891 | Helped |
| Add residual-aware shortlist selection to the inverted multi-index | IMI · SIFT-1B, K=100, shortlist 1M | recall: 0.994 -> 0.992 | Hurt |
| Add residual-aware shortlist selection to the inverted index | II · S-VLAD2K-1B, K=100, shortlist 1M | recall: 0.057 -> 0.082 | Helped |
| Add residual-aware shortlist selection to the inverted multi-index | IMI · S-VLAD2K-1B, K=100, shortlist 1M | recall: 0.158 -> 0.182 | Helped |
| Add the residual-aware estimator to IMI shortlist selection | IMI, VLAD8K-1M · VLAD8K-1M | shortlist computation time: 3.43, 3.54 -> 1.92, 2.75 | Helped |
| Train αK using sampled true neighbors and sampled non-neighbors instead of using default αK=1 (ablation) | Residual-aware distance estimator · Benchmarks tested in the paper | shortlist accuracy: up to 20% | Helped |
| Change the IMI residual-distance partition count from P=2 to P=4 (ablation) | IMI+Ours and OIMI+Ours · Benchmarks tested in the paper | shortlist accuracy | Mixed |
Closest papers by meaning
- Fast k-Nearest Neighbour Search via Dynamic Continuous Indexing (ICML 2016 · similarity 0.87)
- SOAR: Improved Indexing for Approximate Nearest Neighbor Search (NeurIPS 2023 · similarity 0.87)
- Graph-Based Algorithms for Diverse Similarity Search (ICML 2025 · similarity 0.86)
- Distance Adaptive Beam Search for Provably Accurate Graph-Based Nearest Neighbor Search (NeurIPS 2025 · similarity 0.85)
- LotusFilter: Fast Diverse Nearest Neighbor Search via a Learned Cutoff Table (CVPR 2025 · similarity 0.84)
- Weighted distance nearest neighbor condensing (ICML 2024 · 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.