Research map / Learning theory and algorithms / Prediction and decision losses
Non-metric Similarity Graphs for 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 |
|---|---|---|---|
| run similarity graph on original data with inner product instead of after MIPS-to-NNS reduction (ablation) | NSW graph · Music-100, Yahoo! Music, Normal-64 | recall vs time: ip-NSW reaches recall 0.9 five times faster on Music-100 | Helped |
| use inner-product similarity for graph construction instead of reduction plus Euclidean (ablation) | NSW graph · Music-100 | rate of edges to larger-norm vectors: 0.349335 -> 0.75347 | Helped |
| use inner-product similarity for graph construction instead of reduction plus Euclidean (ablation) | NSW graph · Yahoo! Music | rate of edges to larger-norm vectors: 0.398541 -> 0.92353 | Helped |
| use inner-product similarity for graph construction instead of reduction plus Euclidean (ablation) | NSW graph · Normal-64 | rate of edges to larger-norm vectors: 0.362722 -> 0.703605 | Helped |
| use ip-NSW instead of Greedy-MIPS, LEMP, LSH, clustering, FEXIPRO baselines | ip-NSW · Music-100, Yahoo! Music, Normal-64 (top-1/top-10) | recall vs time: 0.9 recall ten times faster than fastest baseline | Helped |
Closest papers by meaning
- Möbius Transformation for Fast Inner Product Search on Graph (NeurIPS 2019 · similarity 0.89)
- Faster Maximum Inner Product Search in High Dimensions (ICML 2024 · similarity 0.84)
- Optimistic Query Routing in Clustering-based Approximate Maximum Inner Product Search (NeurIPS 2025 · similarity 0.84)
- Graph-Based Algorithms for Diverse Similarity Search (ICML 2025 · similarity 0.83)
- Position: Graph Matching Systems Deserve Better Benchmarks (ICML 2025 · similarity 0.83)
- Distance Adaptive Beam Search for Provably Accurate Graph-Based Nearest Neighbor Search (NeurIPS 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.