Research map / Learning theory and algorithms / Prediction and decision losses
Product Quantized Translation for Fast Nearest Neighbor 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 |
|---|---|---|---|
| increase number of clusters from 32 to 128 at partition dimension 16 (ablation) | PQT · MS-COCO | query time (s): 10.78 -> 18.32 | Hurt |
| increase number of clusters from 32 to 128 at partition dimension 32 (ablation) | PQT · SIFT5M | query time (s): 17.46 -> 14.60 | Helped |
| increase number of clusters from 32 to 128 at partition dimension 32 (ablation) | PQT · GIST1M | query time (s): 7.94 -> 9.07 | Hurt |
| increase partition dimension from 16 to 32 at 64 clusters (ablation) | PQT · MNIST | query time (s): 0.55 -> 0.35 | Helped |
| increase partition dimension from 32 to 64 at 64 clusters (ablation) | PQT · GIST1M | query time (s): 8.39 -> 10.45 | Hurt |
| apply k-means product quantized translation to data and query vs no translation (ablation) | PQT lower bound · MNIST, SIFT5M, GIST1M, MS-COCO | lower bound quality LB/||x-y||^2 | Helped |
| k-means centroids instead of random centroids for translations (ablation) | PQT lower bound · four datasets | lower bound quality | Helped |
| overall PQT parameter variation (ablation) | PQT · four datasets | query time (s) | No clear effect |
Closest papers by meaning
- Efficient Large-Scale Approximate Nearest Neighbor Search on the GPU (CVPR 2016 · similarity 0.88)
- Efficient Large-Scale Approximate Nearest Neighbor Search on OpenCL FPGA (CVPR 2018 · similarity 0.87)
- An algorithm for L1 nearest neighbor search via monotonic embedding (NeurIPS 2016 · similarity 0.87)
- Graph-Based Algorithms for Diverse Similarity Search (ICML 2025 · similarity 0.85)
- LotusFilter: Fast Diverse Nearest Neighbor Search via a Learned Cutoff Table (CVPR 2025 · similarity 0.85)
- Approximate Nearest Neighbor Search with Window Filters (ICML 2024 · 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.