Research map / Learning theory and algorithms / Prediction and decision losses
Rejection Sampling for Weighted Jaccard Similarity Revisited
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 early stopping and densification with tabulated random sequences (ERS) instead of RS | Weighted Jaccard hashing · Synthetic data (d=4096, K=256) | running time (ms): 0.80 -> 0.06 | Helped |
| Use early stopping and densification with tabulated random sequences (ERS) instead of DartMinHash | Weighted Jaccard hashing · Synthetic data (d=16384, K=256) | running time (ms): 2.79 -> 0.03 | Helped |
| Use early stopping and densification with tabulated random sequences (ERS) instead of CWS | Weighted Jaccard hashing · Synthetic data (d=4096, K=256) | running time (ms): 18.05 -> 0.06 | Helped |
| Increase sequence length from L=0.5/s to L=5/s (ablation) | ERS · Synthetic data (d=4096, K=256) | running time (ms): 0.06 -> 0.14 | Hurt |
| Increase sequence length from L=0.5/s to L=1/s (ablation) | ERS · Synthetic data (d=4096, K=256) | running time (ms): 0.06 -> 0.08 | Hurt |
| Increase sequence length L (ablation) | ERS Jaccard estimator · Weighted Jaccard estimation | estimation bias | Helped |
| Use a very small sequence length L (ablation) | ERS Jaccard estimator · Words and Caltech101 | mean absolute error | Hurt |
| Increase L from 100 to 300 or 500 (ablation) | ERS kernel-based SVM · Dailysports and PCMAC | test accuracy | Helped |
| Use L=100–300 rather than larger L (ablation) | ERS kernel-based SVM · Dailysports | test accuracy | No clear effect |
| Use L=300 (ablation) | ERS kernel-based SVM · PCMAC | test accuracy | No clear effect |
Closest papers by meaning
- Channel Simulation and Distributed Compression with Ensemble Rejection Sampling (NeurIPS 2025 · similarity 0.82)
- A Separation in Heavy-Tailed Sampling: Gaussian vs. Stable Oracles for Proximal Samplers (NeurIPS 2024 · similarity 0.82)
- Query-Efficient Correlation Clustering with Noisy Oracle (NeurIPS 2024 · similarity 0.81)
- Weak-to-Strong Generalization Through the Data-Centric Lens (ICLR 2025 · similarity 0.81)
- Boosting Test Performance with Importance Sampling--a Subpopulation Perspective (AAAI 2025 · similarity 0.81)
- Data-Efficient Learning via Clustering-Based Sensitivity Sampling: Foundation Models and Beyond (ICML 2024 · similarity 0.81)
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.