Research map / Learning theory and algorithms / Prediction and decision losses
SAH: Shifting-Aware Asymmetric Hashing for Reverse k Maximum Inner Product Search
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 SAH instead of Simpfer | RkMIPS · Five recommendation datasets | query time: 4~8× faster | Helped |
| Use the shifting-invariant transformation in SA-ALSH | SAH · Five recommendation datasets | F1-score | Helped |
| Use kMIPS results directly to answer RkMIPS | RkMIPS · Amazon-Auto, Amazon-CDs, MovieLens, Music100, and Netflix | F1-score: 0.0, 0.0, 0.1, 0.1, 0.1; 0.2, 0.4, 0.7, 1.3, 2.8; 0.0, 26.9, 25.9, 10.5, 10.3; 11.1, 31.7, 30.6, 21.7, 18.6; 5.1, 5.7, 7 | Hurt |
| Add SA-ALSH rather than using H2-ALSH (ablation) | RkMIPS · Five recommendation datasets | query time | Helped |
| Add SA-ALSH rather than using H2-ALSH (ablation) | RkMIPS · Five recommendation datasets | F1-score | Helped |
| Add Cone-Tree blocking rather than norm-based blocking (ablation) | RkMIPS · Five recommendation datasets | query time | Helped |
| Combine SA-ALSH and Cone-Tree blocking (ablation) | RkMIPS · Five recommendation datasets | query time | Helped |
| Increase SA-ALSH hash tables from 64 to 128 (ablation) | SAH · Recommendation datasets, k=10 | F1-score | Helped |
| Increase SA-ALSH hash tables (ablation) | SAH · Recommendation datasets, k=10 | query time | Hurt |
| Vary Cone-Tree maximum leaf size (ablation) | SAH · Recommendation datasets, k=10 | query time | No clear effect |
| Set interval ratio b to 0.5 (ablation) | SAH · Recommendation datasets, k=10 | efficiency-accuracy trade-off | Helped |
Closest papers by meaning
- Asymmetric Deep Supervised Hashing (AAAI 2018 · similarity 0.87)
- Norm-Ranging LSH for Maximum Inner Product Search (NeurIPS 2018 · similarity 0.87)
- Sublinear Time Nearest Neighbor Search over Generalized Weighted Space (ICML 2019 · similarity 0.87)
- AMVH: Asymmetric Multi-Valued Hashing (CVPR 2017 · similarity 0.86)
- Falconn++: A Locality-sensitive Filtering Approach for Approximate Nearest Neighbor Search (NeurIPS 2022 · similarity 0.85)
- KALAHash: Knowledge-Anchored Low-Resource Adaptation for Deep Hashing (AAAI 2025 · similarity 0.83)
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.