Research map / Architectures and efficiency / Quantization
Product Quantizer Aware Inverted Index for Scalable Nearest Neighbor Search
Haechan Noh, Taeho Kim, Jae-Pil Heo
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 joint coarse-fine optimization (ablation) | IVFADC · SIFT-1M 64bit | recall@1: 0.2962 -> 0.3108 | Improved |
| add joint coarse-fine optimization (ablation) | IVFOADC · SIFT-1M 64bit | recall@1: 0.2952 -> 0.3214 | Improved |
| add joint coarse-fine optimization (ablation) | Multi-D-ADC · SIFT-1M 64bit | recall@1: 0.3027 -> 0.3181 | Improved |
| add joint coarse-fine optimization (ablation) | IVFADC · SIFT-1M 128bit | recall@1: 0.4714 -> 0.4964 | Improved |
| add joint coarse-fine optimization (ablation) | IVFADC · SIFT-1B | recall@1: 0.0973 -> 0.1143 | Improved |
| add joint coarse-fine optimization (ablation) | IVFADC · SIFT-1M | recall@1: 4.93% | Improved |
| add joint coarse-fine optimization (ablation) | IVFADC · Deep-1B | recall@1: 0.1738 -> 0.1751 | Improved |
| add joint coarse-fine optimization (ablation) | Multi-D-ADC K'=2^10 · Deep-1B | recall@10: 0.3538 -> 0.3529 | No clear effect |
| add joint optimization at large candidate set L=10M (ablation) | IVFADC · SIFT-1B | recall@100: 7.65% | Improved |
| add joint optimization at small candidate set L=0.1M (ablation) | IVFADC · SIFT-1B | recall@100: 1.93% | Improved |
| add joint optimization on OPQ vs using OPQ alone (ablation) | IVFADC · Conv-1M 64bit | recall@10: 3.77% | Improved |
| add joint optimization over coarse index sizes K' from 2^9 to 2^11 (ablation) | IVFADC · SIFT-1M 64bit | recall@1: 0.2838 -> 0.2985 | Improved |
Closest papers by meaning
- Norm-Explicit Quantization: Improving Vector Quantization for Maximum Inner Product Search (AAAI 2020 · similarity 0.90)
- Product Quantized Translation for Fast Nearest Neighbor Search (AAAI 2018 · similarity 0.89)
- Revisiting the Inverted Indices for Billion-Scale Approximate Nearest Neighbors (ECCV 2018 · similarity 0.88)
- Composite Quantization (TPAMI 2018 · similarity 0.88)
- Multiscale Quantization for Fast Similarity Search (NeurIPS 2017 · similarity 0.88)
- Qinco2: Vector Compression and Search with Improved Implicit Neural Codebooks (ICLR 2025 · 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.
